Update libkv dependency

cluster/datastore.go

@@ -76,7 +76,7 @@ func NewDataStore(ctx context.Context, kvSource staert.KvSource, object Object,
 
 func (d *Datastore) watchChanges() error {
 	stopCh := make(chan struct{})
-	kvCh, err := d.kv.Watch(d.lockKey, stopCh)
+	kvCh, err := d.kv.Watch(d.lockKey, stopCh, nil)
 	if err != nil {
 		return err
 	}

docs/configuration/backends/etcd.md

@@ -31,6 +31,16 @@ watch = true
 #
 prefix = "/traefik"
 
+# Force to use API V3 (otherwise still use API V2)
+#
+# Deprecated
+#
+# Optional
+# Default: false
+#
+useAPIV3 = true
+
+
 # Override default configuration template.
 # For advanced users :)
 #
@@ -59,3 +69,7 @@ prefix = "/traefik"
 To enable constraints see [backend-specific constraints section](/configuration/commons/#backend-specific).
 
 Please refer to the [Key Value storage structure](/user-guide/kv-config/#key-value-storage-structure) section to get documentation on Traefik KV structure.
+
+!!! note
+    The option `useAPIV3` allows using Etcd API V3 only if it's set to true.
+    This option is **deprecated** and API V2 won't be supported in the future.

docs/user-guide/kv-config.md

@@ -322,6 +322,10 @@ As a result, it may be possible for Træfik to read an intermediate configuratio
 To solve this problem, Træfik supports a special key called `/traefik/alias`.
 If set, Træfik use the value as an alternative key prefix.
 
+!!! note
+    The field `useAPIV3` allows using Etcd V3 API which should support updating multiple keys atomically with Etcd.
+    Etcd API V2 is deprecated and, in the future, Træfik will support API V3 by default.
+
 Given the key structure below, Træfik will use the `http://172.17.0.2:80` as its only backend (frontend keys have been omitted for brevity).
 
 | Key                                                                     | Value                       |

examples/cluster/docker-compose.yml

@@ -0,0 +1,210 @@
+version: '2'
+
+services:
+
+  ## KV part ##
+
+  # CONSUL
+
+  consul:
+      image: progrium/consul
+      command: -server -bootstrap -log-level debug -ui-dir /ui
+      ports:
+        - "8400:8400"
+        - "8500:8500"
+        - "8600:53/udp"
+      expose:
+        - "8300"
+        - "8301"
+        - "8301/udp"
+        - "8302"
+        - "8302/udp"
+      networks:
+        net:
+          ipv4_address: 10.0.1.2
+
+  # ETCD V3
+
+  etcd3:
+      image: quay.io/coreos/etcd:v3.2.9
+      command: /usr/local/bin/etcd --data-dir=/etcd-data --name node1 --initial-advertise-peer-urls http://10.0.1.12:2380 --listen-peer-urls http://10.0.1.12:2380 --advertise-client-urls http://10.0.1.12:2379,http://10.0.1.12:4001 --listen-client-urls http://10.0.1.12:2379,http://10.0.1.12:4001  --initial-cluster node1=http://10.0.1.12:2380 --debug
+      ports:
+        - "4001:4001"
+        - "2380:2380"
+        - "2379:2379"
+      networks:
+        net:
+          ipv4_address: 10.0.1.12
+
+  etcdctl-ping:
+      image: tenstartups/etcdctl
+      command: --endpoints=[10.0.1.12:2379] get "traefik/acme/storagefile"
+      environment:
+          ETCDCTL_DIAL_: "TIMEOUT 10s"
+          ETCDCTL_API : "3"
+      networks:
+        - net
+
+  etcdctl-rm:
+      image: tenstartups/etcdctl
+      command: --endpoints=[10.0.1.12:2379] del "/traefik/acme/storagefile"
+      environment:
+          ETCDCTL_DIAL_: "TIMEOUT 10s"
+          ETCDCTL_API : "3"
+      networks:
+        - net
+
+  ## BOULDER part ##
+
+  boulder:
+      image: containous/boulder:release
+      environment:
+          FAKE_DNS: 172.17.0.1
+          PKCS11_PROXY_SOCKET: tcp://boulder-hsm:5657
+      extra_hosts:
+        - le.wtf:127.0.0.1
+        - boulder:127.0.0.1
+      ports:
+        - 4000:4000 # ACME
+        - 4002:4002 # OCSP
+        - 4003:4003 # OCSP
+        - 4500:4500 # ct-test-srv
+        - 8000:8000 # debug ports
+        - 8001:8001
+        - 8002:8002
+        - 8003:8003
+        - 8004:8004
+        - 8055:8055 # dns-test-srv updates
+        - 9380:9380 # mail-test-srv
+        - 9381:9381 # mail-test-srv
+      restart: unless-stopped
+      depends_on:
+        - bhsm
+        - bmysql
+        - brabbitmq
+      networks:
+        net:
+          ipv4_address: 10.0.1.3
+
+  bhsm:
+      image: letsencrypt/boulder-tools:2016-11-02
+      hostname: boulder-hsm
+      environment:
+          PKCS11_DAEMON_SOCKET: tcp://0.0.0.0:5657
+      command: /usr/local/bin/pkcs11-daemon /usr/lib/softhsm/libsofthsm.so
+      expose:
+        - 5657
+      networks:
+        net:
+          ipv4_address: 10.0.1.4
+          aliases:
+            - boulder-hsm
+  bmysql:
+      image: mariadb:10.1
+      hostname: boulder-mysql
+      environment:
+          MYSQL_ALLOW_EMPTY_PASSWORD: "yes"
+      networks:
+        net:
+          ipv4_address: 10.0.1.5
+          aliases:
+            - boulder-mysql
+
+  brabbitmq:
+      image: rabbitmq:3-alpine
+      hostname: boulder-rabbitmq
+      environment:
+          RABBITMQ_NODE_IP_ADDRESS: "0.0.0.0"
+      networks:
+        net:
+          ipv4_address: 10.0.1.6
+          aliases:
+            - boulder-rabbitmq
+
+  ## TRAEFIK part ##
+
+  traefik-storeconfig:
+      build:
+        context: ../..
+      image: containous/traefik
+      volumes:
+        - "./traefik.toml:/traefik.toml:ro"
+        - "./acme.json:/acme.json:ro"
+      command: storeconfig --debug
+      networks:
+        - net
+
+  traefik01:
+      build:
+        context: ../..
+      image: containous/traefik
+      command: ${TRAEFIK_CMD}
+      extra_hosts:
+        - traefik.boulder.com:172.17.0.1
+      volumes:
+        - "/var/run/docker.sock:/var/run/docker.sock:ro"
+      expose:
+        - "443"
+        - "5001"
+      ports:
+        - "80:80"
+        - "8080:8080"
+        - "443:443"
+        - "5001:443" # Needed for SNI challenge
+      networks:
+        net:
+          ipv4_address: 10.0.1.8
+
+  traefik02:
+      build:
+        context: ../..
+      image: containous/traefik
+      command: ${TRAEFIK_CMD}
+      extra_hosts:
+        - traefik.boulder.com:172.17.0.1
+      volumes:
+        - "/var/run/docker.sock:/var/run/docker.sock:ro"
+      expose:
+        - "443"
+        - "5001"
+      ports:
+        - "88:80"
+        - "8888:8080"
+        - "8443:443"
+      depends_on:
+        - traefik01
+      networks:
+        net:
+          ipv4_address: 10.0.1.9
+
+  whoami01:
+      image: emilevauge/whoami
+      expose:
+        - "80"
+      labels:
+        - "traefik.port=80"
+        - "traefik.backend=wam01"
+        - "traefik.frontend.rule=Host:who01.localhost.com"
+        - "traefik.enable=true"
+      networks:
+        net:
+          ipv4_address: 10.0.1.10
+
+  whoami02:
+      image: emilevauge/whoami
+      expose:
+        - "80"
+      labels:
+        - "traefik.port=80"
+        - "traefik.backend=wam02"
+        - "traefik.frontend.rule=Host:who02.localhost.com"
+        - "traefik.enable=true"
+      networks:
+        - net
+
+networks:
+  net:
+    driver: bridge
+    ipam:
+     config:
+       - subnet: 10.0.1.0/26

examples/cluster/manage_cluster_docker_environment.sh

@@ -0,0 +1,247 @@
+#! /usr/bin/env bash
+
+# Initialize variables
+readonly basedir=$(dirname $0)
+readonly doc_file=$basedir"/docker-compose.yml"
+export COMPOSE_PROJECT_NAME="cluster"
+
+# Stop and remove Docker environment
+down_environment() {
+    echo "DOWN Docker environment"
+    ! docker-compose -f $doc_file down -v &>/dev/null && \
+        echo "[ERROR] Unable to stop the Docker environment" && exit 11
+    return 0
+}
+
+# Create and start Docker-compose environment or subpart of its services (if services are listed)
+# $@ : List of services to start (optional)
+up_environment() {
+    echo "START Docker environment "$@
+    ! docker-compose -f $doc_file up -d $@ &>/dev/null && \
+        echo "[ERROR] Unable to start Docker environment ${@}" && exit 21
+    return 0
+}
+
+# Stop and remove Docker environment
+delete_services() {
+    echo "DELETE services "$@
+    ! docker-compose -f $doc_file stop $@ &>/dev/null && \
+        echo "[ERROR] Unable to stop services "$@ && exit 31
+    ! docker-compose -f $doc_file rm -vf $@ &>/dev/null && \
+        echo "[ERROR] Unable to delete services "$@ && exit 31
+    return 0
+}
+
+# Init the environment : get IP address and create needed files
+init_acme_json() {
+    echo "CREATE empty acme.json file"
+    rm -f $basedir/acme.json && \
+    touch $basedir/acme.json && \
+    echo "{}" > $basedir/acme.json && \
+    chmod 600 $basedir/acme.json # Needed for ACME
+}
+
+start_consul() {
+    up_environment consul
+    waiting_counter=12
+    # Not start Traefik store config if consul is not started
+    echo "WAIT for consul..."
+    sleep 5
+    while [[ -z $(curl -s http://10.0.1.2:8500/v1/status/leader) ]]; do
+        sleep 5
+        let waiting_counter-=1
+        if [[ $waiting_counter -eq 0 ]]; then
+            echo "[ERROR] Unable to start consul container in the allowed time, the Docker environment will be stopped"
+            down_environment
+            exit 41
+        fi
+    done
+
+}
+
+start_etcd3() {
+    up_environment etcd3
+    waiting_counter=12
+    # Not start Traefik store config if consul is not started
+    echo "WAIT for ETCD3..."
+    while [[ -z $(curl -s --connect-timeout 2 http://10.0.1.12:2379/version) ]]; do
+        sleep 5
+        let waiting_counter-=1
+        if [[ $waiting_counter -eq 0 ]]; then
+            echo "[ERROR] Unable to start etcd3 container in the allowed time, the Docker environment will be stopped"
+            down_environment
+            exit 51
+        fi
+    done
+}
+
+start_storeconfig_consul() {
+    init_acme_json
+    # Create traefik.toml with consul provider
+    cp $basedir/traefik.toml.tmpl $basedir/traefik.toml
+    echo '
+        [consul]
+        endpoint = "10.0.1.2:8500"
+        watch = true
+        prefix = "traefik"' >> $basedir/traefik.toml
+    up_environment traefik-storeconfig
+    rm -f $basedir/traefik.toml && rm -f $basedir/acme.json
+    # Delete acme-storage-file key
+    waiting_counter=5
+    # Not start Traefik store config if consul is not started
+    echo "Delete storage file key..."
+    while [[ -z $(curl -s http://10.0.1.2:8500/v1/kv/traefik/acme/storagefile)  &&  $waiting_counter -gt 0 ]]; do
+        sleep 5
+        let waiting_counter-=1
+    done
+    if [[ $waiting_counter -eq 0 ]]; then
+        echo "[WARN] Unable to get storagefile key in consul"
+    else
+        curl -s --request DELETE http://10.0.1.2:8500/v1/kv/traefik/acme/storagefile
+        ret=$1
+        if [[ $ret -ne 0 ]]; then
+            echo "[ERROR] Unable to delete storagefile key from consul kv."
+        fi
+    fi
+
+}
+
+start_storeconfig_etcd3() {
+    init_acme_json
+    # Create traefik.toml with consul provider
+    cp $basedir/traefik.toml.tmpl $basedir/traefik.toml
+    echo '
+        [etcd]
+        endpoint = "10.0.1.12:2379"
+        watch = true
+        prefix = "/traefik"
+        useAPIV3 = true' >> $basedir/traefik.toml
+    up_environment traefik-storeconfig
+    rm -f $basedir/traefik.toml && rm -f $basedir/acme.json
+    # Delete acme-storage-file key
+    waiting_counter=5
+    # Not start Traefik store config if consul is not started
+    echo "Delete storage file key..."
+    while [[ $(docker-compose -f $doc_file up --exit-code-from etcdctl-ping etcdctl-ping &>/dev/null) -ne 0 &&  $waiting_counter -gt 0 ]]; do
+        sleep 5
+        let waiting_counter-=1
+    done
+    # Not start Traefik store config if consul is not started
+    echo "Delete storage file key from ETCD3..."
+
+    up_environment etcdctl-rm && \
+    delete_services etcdctl-rm traefik-storeconfig etcdctl-ping
+}
+
+start_traefik() {
+    up_environment traefik01
+    # Waiting for the first instance which is mapped to the host as leader before to start the second one
+    waiting_counter=5
+    echo "WAIT for traefik leader..."
+    sleep 10
+    while [[ -z $(curl -s --connect-timeout 3 http://10.0.1.8:8080/ping) ]]; do
+        sleep 2
+        let waiting_counter-=1
+        if [[ $waiting_counter -eq 0 ]]; then
+            echo "[ERROR] Unable to start Traefik leader container in the allowed time, the Docker environment will be stopped"
+            down_environment
+            exit 51
+        fi
+    done
+    up_environment whoami01
+    waiting_counter=5
+    echo "WAIT for whoami..."
+    sleep 10
+    while [[ -z $(curl -s --connect-timeout 3 http://10.0.1.10) ]]; do
+        sleep 2
+        let waiting_counter-=1
+        if [[ $waiting_counter -eq 0 ]]; then
+            echo "[ERROR] Unable to start whoami container in the allowed time, the Docker environment will be stopped"
+            down_environment
+            exit 52
+        fi
+    done
+    up_environment traefik02 whoami02
+}
+
+# Start boulder services
+start_boulder() {
+    echo "Start boulder environment"
+    up_environment bmysql brabbitmq bhsm boulder
+    waiting_counter=12
+    # Not start Traefik if boulder is not started
+    echo "WAIT for boulder..."
+    while [[ -z $(curl -s http://10.0.1.3:4000/directory) ]]; do
+        sleep 5
+        let waiting_counter-=1
+        if [[ $waiting_counter -eq 0 ]]; then
+            echo "[ERROR] Unable to start boulder container in the allowed time, the Docker environment will be stopped"
+            down_environment
+            exit 61
+        fi
+    done
+    echo "Boulder started."
+}
+
+# Script usage
+show_usage() {
+    echo
+    echo "USAGE : manage_cluster_docker_environment.sh [--start [--consul|--etcd3]|--stop|--restart [--consul|--etcd3]]"
+    echo
+}
+
+# Main method
+# $@ All parameters given
+main() {
+
+    [[ $# -lt 1 && $# -gt 2 ]] && show_usage && exit 1
+
+    case $1 in
+        "--start")
+                [[ $# -ne 2 ]] && show_usage && exit 2
+                # The domains who01.localhost.com and who02.localhost.com have to refer 127.0.0.1
+                # I, the /etc/hosts file
+                for whoami_idx in "01" "02"; do
+                    [[ -z $(cat /etc/hosts | grep "127.0.0.1" | grep -vE "^#" | grep "who${whoami_idx}.localhost.com") ]] && \
+                        echo "[ERROR] Domain who${whoami_idx}.localhost.com has to refer to 127.0.0.1 into /etc/hosts file." && \
+                        exit 3
+                done
+                case $2 in
+                    "--etcd3")
+                        echo "USE ETCD V3 AS KV STORE"
+                        export TRAEFIK_CMD="--etcd --etcd.endpoint=10.0.1.12:2379 --etcd.useAPIV3=true"
+                        start_boulder && \
+                        start_etcd3 && \
+                        start_storeconfig_etcd3 && \
+                        start_traefik
+                        ;;
+                    "--consul")
+                        echo "USE CONSUL AS KV STORE"
+                        export TRAEFIK_CMD="--consul --consul.endpoint=10.0.1.2:8500"
+                        start_boulder && \
+                        start_consul && \
+                        start_storeconfig_consul && \
+                        start_traefik
+                        ;;
+                    *)
+                        show_usage && exit 4
+                        ;;
+                esac
+            echo "ENVIRONMENT SUCCESSFULLY STARTED"
+            ;;
+        "--stop")
+            ! down_environment
+            echo "ENVIRONMENT SUCCESSFULLY STOPPED"
+            ;;
+        "--restart")
+            [[ $# -ne 2 ]] && show_usage && exit 5
+            down_environment
+            main --start $2
+            ;;
+        *)
+            show_usage && exit 6
+            ;;
+    esac
+}
+
+main $@

examples/cluster/traefik.toml.tmpl

@@ -0,0 +1,28 @@
+logLevel = "DEBUG"
+
+defaultEntryPoints = ["http", "https"]
+
+[entryPoints]
+  [entryPoints.http]
+  address = ":80"
+  [entryPoints.https]
+  address = ":443"
+    [entryPoints.https.tls]
+
+[acme]
+email = "test@traefik.io"
+storage = "traefik/acme/account"
+storageFile = "/acme.json"
+entryPoint = "https"
+OnHostRule = true
+caServer = "http://traefik.boulder.com:4000/directory"
+
+
+[web]
+address = ":8080"
+
+[docker]
+endpoint = "unix:///var/run/docker.sock"
+domain = "localhost.com"
+watch = true
+exposedbydefault = false

examples/etcd-config.sh

@@ -1,25 +1,96 @@
-#!/bin/sh
+#! /usr/bin/env bash
 
-# backend 1
+#
-curl -i -H "Accept: application/json" -X PUT -d value="NetworkErrorRatio() > 0.5"   http://localhost:2379/v2/keys/traefik/backends/backend1/circuitbreaker/expression
+# Insert data in ETCD V3
-curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.2:80"        http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server1/url
+function insert_etcd2_data() {
-curl -i -H "Accept: application/json" -X PUT -d value="10"                          http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server1/weight
+    # backend 1
-curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.3:80"        http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server2/url
+    curl -i -H "Accept: application/json" -X PUT -d value="NetworkErrorRatio() > 0.5"   http://localhost:2379/v2/keys/traefik/backends/backend1/circuitbreaker/expression
-curl -i -H "Accept: application/json" -X PUT -d value="1"                           http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server2/weight
+    curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.2:80"        http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server1/url
+    curl -i -H "Accept: application/json" -X PUT -d value="10"                          http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server1/weight
+    curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.3:80"        http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server2/url
+    curl -i -H "Accept: application/json" -X PUT -d value="1"                           http://localhost:2379/v2/keys/traefik/backends/backend1/servers/server2/weight
 
-# backend 2
+    # backend 2
-curl -i -H "Accept: application/json" -X PUT -d value="drr"                         http://localhost:2379/v2/keys/traefik/backends/backend2/loadbalancer/method
+    curl -i -H "Accept: application/json" -X PUT -d value="drr"                         http://localhost:2379/v2/keys/traefik/backends/backend2/loadbalancer/method
-curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.4:80"        http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server1/url
+    curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.4:80"        http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server1/url
-curl -i -H "Accept: application/json" -X PUT -d value="1"                           http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server1/weight
+    curl -i -H "Accept: application/json" -X PUT -d value="1"                           http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server1/weight
-curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.5:80"        http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server2/url
+    curl -i -H "Accept: application/json" -X PUT -d value="http://172.17.0.5:80"        http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server2/url
-curl -i -H "Accept: application/json" -X PUT -d value="2"                           http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server2/weight
+    curl -i -H "Accept: application/json" -X PUT -d value="2"                           http://localhost:2379/v2/keys/traefik/backends/backend2/servers/server2/weight
 
-# frontend 1
+    # frontend 1
-curl -i -H "Accept: application/json" -X PUT -d value="backend2"                    http://localhost:2379/v2/keys/traefik/frontends/frontend1/backend
+    curl -i -H "Accept: application/json" -X PUT -d value="backend2"                    http://localhost:2379/v2/keys/traefik/frontends/frontend1/backend
-curl -i -H "Accept: application/json" -X PUT -d value="http"                        http://localhost:2379/v2/keys/traefik/frontends/frontend1/entrypoints
+    curl -i -H "Accept: application/json" -X PUT -d value="http"                        http://localhost:2379/v2/keys/traefik/frontends/frontend1/entrypoints
-curl -i -H "Accept: application/json" -X PUT -d value="Host:test.localhost"         http://localhost:2379/v2/keys/traefik/frontends/frontend1/routes/test_1/rule
+    curl -i -H "Accept: application/json" -X PUT -d value="Host:test.localhost"         http://localhost:2379/v2/keys/traefik/frontends/frontend1/routes/test_1/rule
 
-# frontend 2
+    # frontend 2
-curl -i -H "Accept: application/json" -X PUT -d value="backend1"                    http://localhost:2379/v2/keys/traefik/frontends/frontend2/backend
+    curl -i -H "Accept: application/json" -X PUT -d value="backend1"                    http://localhost:2379/v2/keys/traefik/frontends/frontend2/backend
-curl -i -H "Accept: application/json" -X PUT -d value="http"                        http://localhost:2379/v2/keys/traefik/frontends/frontend2/entrypoints
+    curl -i -H "Accept: application/json" -X PUT -d value="http"                        http://localhost:2379/v2/keys/traefik/frontends/frontend2/entrypoints
-curl -i -H "Accept: application/json" -X PUT -d value="Path:/test"                  http://localhost:2379/v2/keys/traefik/frontends/frontend2/routes/test_2/rule
+    curl -i -H "Accept: application/json" -X PUT -d value="Path:/test"                  http://localhost:2379/v2/keys/traefik/frontends/frontend2/routes/test_2/rule
+
+}
+
+#
+# Insert data in ETCD V3
+# $1 = ECTD IP address
+# Note : This function allows adding data in a ETCD V3 which is directly installed on a host
+#        or in container which binds its port 2379 on a host in the way to allows etcd_client container to access it.
+function insert_etcd3_data() {
+
+    readonly etcd_ip=$1
+    # backend 1
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend1/circuitbreaker/expression" "NetworkErrorRatio() > 0.5"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend1/servers/server1/url" "http://172.17.0.2:80"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend1/servers/server1/weight" "10"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend1/servers/server2/url" "http://172.17.0.3:80"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend1/servers/server2/weight" "1"
+
+    # backend 2
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend2/loadbalancer/method" "drr"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend2/servers/server1/url" "http://172.17.0.4:80"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend2/servers/server1/weight" "1"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend2/servers/server2/url" "http://172.17.0.5:80"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/backends/backend2/servers/server2/weight" "2"
+
+    # frontend 1
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/frontends/frontend1/backend" "backend2"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik//frontends/frontend1/entrypoints" "http"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/frontends/frontend1/routes/test_1/rule" "Host:test.localhost"
+
+    # frontend 2
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/frontends/frontend2/backend" "backend1"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/frontends/frontend2/entrypoints" "http"
+    docker container run --rm -ti -e ETCDCTL_DIAL_="TIMEOUT 10s" -e ETCDCTL_API="3" tenstartups/etcdctl --endpoints=[$etcd_ip:2379] put "/traefik/frontends/frontend2/routes/test_2/rule" "Path:/test"
+}
+
+function show_usage() {
+        echo "USAGE : etcd-config.sh ETCD_API_VERSION [ETCD_IP_ADDRESS]"
+        echo "        ETCD_API_VERSION : Values v2 or V3 (v3 requires ETCD_IP_ADDRESS)"
+        echo "        ETCD_IP_ADDRESS : Host ETCD IP address (not 127.0.0.1)"
+}
+
+function main() {
+    case $# in
+        1)
+            if [[ $1 == "V2" ]]; then
+                insert_etcd2_data
+            else
+                show_usage
+                exit 1
+            fi
+            ;;
+        2)
+            if [[ $1 == "V3" && $2 != "127.0.0.1" && ! -z $(echo $2 | grep -oE "([0-9]+(\.)?){4}") ]]; then
+                insert_etcd3_data $2
+            else
+                show_usage
+                exit 1
+            fi
+            ;;
+        *)
+            show_usage
+            exit 1
+            ;;
+    esac
+}
+
+main $@

glide.lock

@@ -1,5 +1,5 @@
-hash: 6e35766dd261f1eb0c2dfaadb8e0bf235eb4f3f942de776f9b48faf8a868a2d0
+hash: 2a604d8b74e8659df7db72d063432fa0822fdee6c81bc6657efa3c1bf0d9cd8a
-updated: 2017-11-15T18:39:20.364720581+02:00
+updated: 2017-11-17T14:21:55.148450413+01:00
 imports:
 - name: cloud.google.com/go
   version: 2e6a95edb1071d750f6d7db777bf66cd2997af6c
@@ -92,13 +92,27 @@ imports:
 - name: github.com/containous/mux
   version: 06ccd3e75091eb659b1d720cda0e16bc7057954c
 - name: github.com/containous/staert
-  version: 1e26a71803e428fd933f5f9c8e50a26878f53147
+  version: af517d5b70db9c4b0505e0144fcc62b054057d2a
+- name: github.com/coreos/bbolt
+  version: 3c6cbfb299c11444eb2f8c9d48f0d2ce09157423
 - name: github.com/coreos/etcd
-  version: c400d05d0aa73e21e431c16145e558d624098018
+  version: f1d7dd87da3e8feab4aaf675b8e29c6a5ed5f58b
   subpackages:
+  - auth/authpb
   - client
+  - clientv3
+  - clientv3/concurrency
+  - etcdserver/api/v3rpc/rpctypes
+  - etcdserver/etcdserverpb
+  - mvcc/mvccpb
   - pkg/pathutil
+  - pkg/srv
   - pkg/types
+  - version
+- name: github.com/coreos/go-semver
+  version: 8ab6407b697782a06568d4b7f1db25550ec2e4c6
+  subpackages:
+  - semver
 - name: github.com/coreos/go-oidc
   version: 5644a2f50e2d2d5ba0b474bc5bc55fea1925936d
   subpackages:
@@ -202,14 +216,20 @@ imports:
 - name: github.com/docker/go-units
   version: 9e638d38cf6977a37a8ea0078f3ee75a7cdb2dd1
 - name: github.com/docker/leadership
-  version: 0a913e2d71a12fd14a028452435cb71ac8d82cb6
+  version: af20da7d3e62be9259835e93261acf931b5adecf
+  repo: https://github.com/containous/leadership.git
+  vcs: git
 - name: github.com/docker/libkv
-  version: 93ab0e6c056d325dfbb11e1d58a3b4f5f62e7f3c
+  version: 5e4bb288a9a74320bb03f5c18d6bdbab0d8049de
+  repo: https://github.com/abronan/libkv.git
+  vcs: git
   subpackages:
   - store
   - store/boltdb
   - store/consul
   - store/etcd
+  - store/etcd/v2
+  - store/etcd/v3
   - store/zookeeper
 - name: github.com/docker/libtrust
   version: 9cbd2a1374f46905c68a4eb3694a130610adc62a
@@ -278,9 +298,11 @@ imports:
 - name: github.com/golang/glog
   version: 44145f04b68cf362d9c4df2182967c2275eaefed
 - name: github.com/golang/protobuf
-  version: 2bba0603135d7d7f5cb73b2125beeda19c09f4ef
+  version: 4bd1920723d7b7c925de087aa32e2187708897f7
   subpackages:
+  - jsonpb
   - proto
+  - ptypes/any
 - name: github.com/google/go-github
   version: fe7d11f8add400587b6718d9f39a62e42cb04c28
   subpackages:
@@ -552,7 +574,7 @@ imports:
   - pbkdf2
   - scrypt
 - name: golang.org/x/net
-  version: 242b6b35177ec3909636b6cf6a47e8c2c6324b5d
+  version: c8c74377599bd978aee1cf3b9b63a8634051cec2
   subpackages:
   - context
   - context/ctxhttp
@@ -578,9 +600,10 @@ imports:
   - unix
   - windows
 - name: golang.org/x/text
-  version: 2910a502d2bf9e43193af9d68ca516529614eed3
+  version: 4ee4af566555f5fbe026368b75596286a312663a
   subpackages:
   - cases
+  - internal
   - internal/tag
   - language
   - runes
@@ -613,11 +636,16 @@ imports:
   - internal/remote_api
   - internal/urlfetch
   - urlfetch
+- name: google.golang.org/genproto
+  version: 09f6ed296fc66555a25fe4ce95173148778dfa85
+  subpackages:
+  - googleapis/rpc/status
 - name: google.golang.org/grpc
-  version: cdee119ee21e61eef7093a41ba148fa83585e143
+  version: b8669c35455183da6d5c474ea6e72fbf55183274
   subpackages:
   - codes
   - credentials
+  - grpclb/grpc_lb_v1
   - grpclog
   - internal
   - keepalive
@@ -625,6 +653,7 @@ imports:
   - naming
   - peer
   - stats
+  - status
   - tap
   - transport
 - name: gopkg.in/fsnotify.v1

glide.yaml

@@ -26,7 +26,7 @@ import:
 - package: github.com/urfave/negroni
   version: 490e6a555d47ca891a89a150d0c1ef3922dfffe9
 - package: github.com/containous/staert
-  version: 1e26a71803e428fd933f5f9c8e50a26878f53147
+  version: ^v2.0.0
 - package: github.com/docker/docker
   version: 75c7536d2e2e328b644bf69153de879d1d197988
 - package: github.com/docker/go-connections
@@ -36,12 +36,17 @@ import:
   - tlsconfig
 - package: github.com/docker/go-units
   version: 9e638d38cf6977a37a8ea0078f3ee75a7cdb2dd1
+- package: github.com/coreos/etcd
+  version: v3.2.9
 - package: github.com/docker/libkv
+  repo: https://github.com/abronan/libkv.git
+  vcs: git
   subpackages:
   - store
   - store/boltdb
   - store/consul
-  - store/etcd
+  - store/etcd/v2
+  - store/etcd/v3
   - store/zookeeper
 - package: github.com/elazarl/go-bindata-assetfs
 - package: github.com/containous/mux
@@ -83,6 +88,8 @@ import:
   repo: https://github.com/containous/gziphandler.git
   vcs: git
 - package: github.com/docker/leadership
+  repo: https://github.com/containous/leadership.git
+  vcs: git
 - package: github.com/satori/go.uuid
   version: ^1.1.0
 - package: k8s.io/client-go
@@ -187,7 +194,7 @@ import:
 - package: github.com/googleapis/gax-go
   version: 9af46dd5a1713e8b5cd71106287eba3cefdde50b
 - package: google.golang.org/grpc
-  version: v1.2.0
+  version: v1.5.2
 - package: github.com/unrolled/secure
   version: 824e85271811af89640ea25620c67f6c2eed987e
 - package: github.com/Nvveen/Gotty

integration/consul_test.go

@@ -346,7 +346,7 @@ func (s *ConsulSuite) TestCommandStoreConfig(c *check.C) {
 	for key, value := range checkmap {
 		var p *store.KVPair
 		err = try.Do(60*time.Second, func() error {
-			p, err = s.kv.Get(key)
+			p, err = s.kv.Get(key, nil)
 			return err
 		})
 		c.Assert(err, checker.IsNil)

integration/etcd3_test.go

@@ -0,0 +1,428 @@
+package integration
+
+import (
+	"crypto/tls"
+	"io/ioutil"
+	"net/http"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/containous/traefik/integration/try"
+	"github.com/docker/libkv"
+	"github.com/docker/libkv/store"
+	"github.com/docker/libkv/store/etcd/v3"
+	"github.com/go-check/check"
+
+	checker "github.com/vdemeester/shakers"
+)
+
+const (
+	// Services IP addresses fixed in the configuration
+	ipEtcd     string = "172.18.0.2"
+	ipWhoami01 string = "172.18.0.3"
+	ipWhoami02 string = "172.18.0.4"
+	ipWhoami03 string = "172.18.0.5"
+	ipWhoami04 string = "172.18.0.6"
+
+	traefikEtcdURL    string = "http://127.0.0.1:8000/"
+	traefikWebEtcdURL string = "http://127.0.0.1:8081/"
+)
+
+// Etcd test suites (using libcompose)
+type Etcd3Suite struct {
+	BaseSuite
+	kv store.Store
+}
+
+func (s *Etcd3Suite) SetUpTest(c *check.C) {
+	s.createComposeProject(c, "etcd3")
+	s.composeProject.Start(c)
+
+	etcdv3.Register()
+	url := ipEtcd + ":2379"
+	kv, err := libkv.NewStore(
+		store.ETCDV3,
+		[]string{url},
+		&store.Config{
+			ConnectionTimeout: 30 * time.Second,
+		},
+	)
+	if err != nil {
+		c.Fatal("Cannot create store etcd")
+	}
+	s.kv = kv
+
+	// wait for etcd
+	err = try.Do(60*time.Second, func() error {
+		_, err := kv.Exists("test", nil)
+		return err
+	})
+	c.Assert(err, checker.IsNil)
+}
+
+func (s *Etcd3Suite) TearDownTest(c *check.C) {
+	// shutdown and delete compose project
+	if s.composeProject != nil {
+		s.composeProject.Stop(c)
+	}
+}
+
+func (s *Etcd3Suite) TearDownSuite(c *check.C) {}
+
+func (s *Etcd3Suite) TestSimpleConfiguration(c *check.C) {
+	file := s.adaptFile(c, "fixtures/etcd/simple.toml", struct {
+		EtcdHost string
+		UseAPIV3 bool
+	}{
+		ipEtcd,
+		true,
+	})
+	defer os.Remove(file)
+
+	cmd, display := s.traefikCmd(withConfigFile(file))
+	defer display(c)
+	err := cmd.Start()
+	c.Assert(err, checker.IsNil)
+	defer cmd.Process.Kill()
+
+	// TODO validate : run on 80
+	// Expected a 404 as we did not configure anything
+	err = try.GetRequest(traefikEtcdURL, 1*time.Second, try.StatusCodeIs(http.StatusNotFound))
+	c.Assert(err, checker.IsNil)
+}
+
+func (s *Etcd3Suite) TestNominalConfiguration(c *check.C) {
+	file := s.adaptFile(c, "fixtures/etcd/simple.toml", struct {
+		EtcdHost string
+		UseAPIV3 bool
+	}{
+		ipEtcd,
+		true,
+	})
+	defer os.Remove(file)
+
+	cmd, display := s.traefikCmd(withConfigFile(file))
+	defer display(c)
+	err := cmd.Start()
+	c.Assert(err, checker.IsNil)
+	defer cmd.Process.Kill()
+
+	backend1 := map[string]string{
+		"/traefik/backends/backend1/circuitbreaker/expression": "NetworkErrorRatio() > 0.5",
+		"/traefik/backends/backend1/servers/server1/url":       "http://" + ipWhoami01 + ":80",
+		"/traefik/backends/backend1/servers/server1/weight":    "10",
+		"/traefik/backends/backend1/servers/server2/url":       "http://" + ipWhoami02 + ":80",
+		"/traefik/backends/backend1/servers/server2/weight":    "1",
+	}
+	backend2 := map[string]string{
+		"/traefik/backends/backend2/loadbalancer/method":    "drr",
+		"/traefik/backends/backend2/servers/server1/url":    "http://" + ipWhoami03 + ":80",
+		"/traefik/backends/backend2/servers/server1/weight": "1",
+		"/traefik/backends/backend2/servers/server2/url":    "http://" + ipWhoami04 + ":80",
+		"/traefik/backends/backend2/servers/server2/weight": "2",
+	}
+	frontend1 := map[string]string{
+		"/traefik/frontends/frontend1/backend":            "backend2",
+		"/traefik/frontends/frontend1/entrypoints":        "http",
+		"/traefik/frontends/frontend1/priority":           "1",
+		"/traefik/frontends/frontend1/routes/test_1/rule": "Host:test.localhost",
+	}
+	frontend2 := map[string]string{
+		"/traefik/frontends/frontend2/backend":            "backend1",
+		"/traefik/frontends/frontend2/entrypoints":        "http",
+		"/traefik/frontends/frontend2/priority":           "10",
+		"/traefik/frontends/frontend2/routes/test_2/rule": "Path:/test",
+	}
+	for key, value := range backend1 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range backend2 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range frontend1 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range frontend2 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+
+	// wait for etcd
+	err = try.Do(60*time.Second, func() error {
+		_, err := s.kv.Exists("/traefik/frontends/frontend2/routes/test_2/rule", nil)
+		return err
+	})
+	c.Assert(err, checker.IsNil)
+
+	// wait for traefik
+	err = try.GetRequest(traefikWebEtcdURL+"api/providers", 60*time.Second, try.BodyContains("Path:/test"))
+	c.Assert(err, checker.IsNil)
+
+	client := &http.Client{}
+	req, err := http.NewRequest(http.MethodGet, traefikEtcdURL, nil)
+	c.Assert(err, checker.IsNil)
+	req.Host = "test.localhost"
+	response, err := client.Do(req)
+
+	c.Assert(err, checker.IsNil)
+	c.Assert(response.StatusCode, checker.Equals, http.StatusOK)
+
+	body, err := ioutil.ReadAll(response.Body)
+	c.Assert(err, checker.IsNil)
+	if !strings.Contains(string(body), ipWhoami03) &&
+		!strings.Contains(string(body), ipWhoami04) {
+		c.Fail()
+	}
+
+	req, err = http.NewRequest(http.MethodGet, traefikEtcdURL+"test", nil)
+	c.Assert(err, checker.IsNil)
+	response, err = client.Do(req)
+
+	c.Assert(err, checker.IsNil)
+	c.Assert(response.StatusCode, checker.Equals, http.StatusOK)
+
+	body, err = ioutil.ReadAll(response.Body)
+	c.Assert(err, checker.IsNil)
+	if !strings.Contains(string(body), ipWhoami01) &&
+		!strings.Contains(string(body), ipWhoami02) {
+		c.Fail()
+	}
+
+	req, err = http.NewRequest(http.MethodGet, traefikEtcdURL+"test2", nil)
+	c.Assert(err, checker.IsNil)
+	req.Host = "test2.localhost"
+	resp, err := client.Do(req)
+	c.Assert(err, checker.IsNil)
+	c.Assert(resp.StatusCode, checker.Equals, http.StatusNotFound)
+
+	resp, err = http.Get(traefikEtcdURL)
+	c.Assert(err, checker.IsNil)
+	c.Assert(resp.StatusCode, checker.Equals, http.StatusNotFound)
+}
+
+func (s *Etcd3Suite) TestGlobalConfiguration(c *check.C) {
+	err := s.kv.Put("/traefik/entrypoints/http/address", []byte(":8001"), nil)
+	c.Assert(err, checker.IsNil)
+
+	// wait for etcd
+	err = try.Do(60*time.Second, func() error {
+		_, err := s.kv.Exists("/traefik/entrypoints/http/address", nil)
+		return err
+	})
+	c.Assert(err, checker.IsNil)
+
+	// start traefik
+	cmd, display := s.traefikCmd(
+		withConfigFile("fixtures/simple_web.toml"),
+		"--etcd",
+		"--etcd.endpoint="+ipEtcd+":4001",
+		"--etcd.useAPIV3=true")
+	defer display(c)
+	err = cmd.Start()
+	c.Assert(err, checker.IsNil)
+	defer cmd.Process.Kill()
+
+	backend1 := map[string]string{
+		"/traefik/backends/backend1/circuitbreaker/expression": "NetworkErrorRatio() > 0.5",
+		"/traefik/backends/backend1/servers/server1/url":       "http://" + ipWhoami01 + ":80",
+		"/traefik/backends/backend1/servers/server1/weight":    "10",
+		"/traefik/backends/backend1/servers/server2/url":       "http://" + ipWhoami02 + ":80",
+		"/traefik/backends/backend1/servers/server2/weight":    "1",
+	}
+	backend2 := map[string]string{
+		"/traefik/backends/backend2/loadbalancer/method":    "drr",
+		"/traefik/backends/backend2/servers/server1/url":    "http://" + ipWhoami03 + ":80",
+		"/traefik/backends/backend2/servers/server1/weight": "1",
+		"/traefik/backends/backend2/servers/server2/url":    "http://" + ipWhoami04 + ":80",
+		"/traefik/backends/backend2/servers/server2/weight": "2",
+	}
+	frontend1 := map[string]string{
+		"/traefik/frontends/frontend1/backend":            "backend2",
+		"/traefik/frontends/frontend1/entrypoints":        "http",
+		"/traefik/frontends/frontend1/priority":           "1",
+		"/traefik/frontends/frontend1/routes/test_1/rule": "Host:test.localhost",
+	}
+	frontend2 := map[string]string{
+		"/traefik/frontends/frontend2/backend":            "backend1",
+		"/traefik/frontends/frontend2/entrypoints":        "http",
+		"/traefik/frontends/frontend2/priority":           "10",
+		"/traefik/frontends/frontend2/routes/test_2/rule": "Path:/test",
+	}
+	for key, value := range backend1 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range backend2 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range frontend1 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range frontend2 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+
+	// wait for etcd
+	err = try.Do(60*time.Second, func() error {
+		_, err := s.kv.Exists("/traefik/frontends/frontend2/routes/test_2/rule", nil)
+		return err
+	})
+	c.Assert(err, checker.IsNil)
+
+	// wait for traefik
+	err = try.GetRequest("http://127.0.0.1:8080/api/providers", 60*time.Second, try.BodyContains("Path:/test"))
+	c.Assert(err, checker.IsNil)
+
+	//check
+	req, err := http.NewRequest(http.MethodGet, "http://127.0.0.1:8001/", nil)
+	c.Assert(err, checker.IsNil)
+	req.Host = "test.localhost"
+
+	err = try.Request(req, 500*time.Millisecond, try.StatusCodeIs(http.StatusOK))
+	c.Assert(err, checker.IsNil)
+}
+
+func (s *Etcd3Suite) TestCertificatesContentstWithSNIConfigHandshake(c *check.C) {
+	// start traefik
+	cmd, display := s.traefikCmd(
+		withConfigFile("fixtures/simple_web.toml"),
+		"--etcd",
+		"--etcd.endpoint="+ipEtcd+":4001",
+		"--etcd.useAPIV3=true")
+	defer display(c)
+
+	//Copy the contents of the certificate files into ETCD
+	snitestComCert, err := ioutil.ReadFile("fixtures/https/snitest.com.cert")
+	c.Assert(err, checker.IsNil)
+	snitestComKey, err := ioutil.ReadFile("fixtures/https/snitest.com.key")
+	c.Assert(err, checker.IsNil)
+	snitestOrgCert, err := ioutil.ReadFile("fixtures/https/snitest.org.cert")
+	c.Assert(err, checker.IsNil)
+	snitestOrgKey, err := ioutil.ReadFile("fixtures/https/snitest.org.key")
+	c.Assert(err, checker.IsNil)
+
+	globalConfig := map[string]string{
+		"/traefik/entrypoints/https/address":                     ":4443",
+		"/traefik/entrypoints/https/tls/certificates/0/certfile": string(snitestComCert),
+		"/traefik/entrypoints/https/tls/certificates/0/keyfile":  string(snitestComKey),
+		"/traefik/entrypoints/https/tls/certificates/1/certfile": string(snitestOrgCert),
+		"/traefik/entrypoints/https/tls/certificates/1/keyfile":  string(snitestOrgKey),
+		"/traefik/defaultentrypoints/0":                          "https",
+	}
+
+	backend1 := map[string]string{
+		"/traefik/backends/backend1/circuitbreaker/expression": "NetworkErrorRatio() > 0.5",
+		"/traefik/backends/backend1/servers/server1/url":       "http://" + ipWhoami01 + ":80",
+		"/traefik/backends/backend1/servers/server1/weight":    "10",
+		"/traefik/backends/backend1/servers/server2/url":       "http://" + ipWhoami02 + ":80",
+		"/traefik/backends/backend1/servers/server2/weight":    "1",
+	}
+	backend2 := map[string]string{
+		"/traefik/backends/backend2/loadbalancer/method":    "drr",
+		"/traefik/backends/backend2/servers/server1/url":    "http://" + ipWhoami03 + ":80",
+		"/traefik/backends/backend2/servers/server1/weight": "1",
+		"/traefik/backends/backend2/servers/server2/url":    "http://" + ipWhoami04 + ":80",
+		"/traefik/backends/backend2/servers/server2/weight": "2",
+	}
+	frontend1 := map[string]string{
+		"/traefik/frontends/frontend1/backend":            "backend2",
+		"/traefik/frontends/frontend1/entrypoints":        "http",
+		"/traefik/frontends/frontend1/priority":           "1",
+		"/traefik/frontends/frontend1/routes/test_1/rule": "Host:snitest.com",
+	}
+	frontend2 := map[string]string{
+		"/traefik/frontends/frontend2/backend":            "backend1",
+		"/traefik/frontends/frontend2/entrypoints":        "http",
+		"/traefik/frontends/frontend2/priority":           "10",
+		"/traefik/frontends/frontend2/routes/test_2/rule": "Host:snitest.org",
+	}
+	for key, value := range globalConfig {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range backend1 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range backend2 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range frontend1 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+	for key, value := range frontend2 {
+		err := s.kv.Put(key, []byte(value), nil)
+		c.Assert(err, checker.IsNil)
+	}
+
+	// wait for etcd
+	err = try.Do(60*time.Second, try.KVExists(s.kv, "/traefik/frontends/frontend1/backend"))
+	c.Assert(err, checker.IsNil)
+
+	err = cmd.Start()
+	c.Assert(err, checker.IsNil)
+	defer cmd.Process.Kill()
+
+	// wait for traefik
+	err = try.GetRequest("http://127.0.0.1:8080/api/providers", 60*time.Second, try.BodyContains("Host:snitest.org"))
+	c.Assert(err, checker.IsNil)
+
+	//check
+	tlsConfig := &tls.Config{
+		InsecureSkipVerify: true,
+		ServerName:         "snitest.com",
+	}
+	conn, err := tls.Dial("tcp", "127.0.0.1:4443", tlsConfig)
+	c.Assert(err, checker.IsNil, check.Commentf("failed to connect to server"))
+
+	defer conn.Close()
+	err = conn.Handshake()
+	c.Assert(err, checker.IsNil, check.Commentf("TLS handshake error"))
+
+	cs := conn.ConnectionState()
+	err = cs.PeerCertificates[0].VerifyHostname("snitest.com")
+	c.Assert(err, checker.IsNil, check.Commentf("certificate did not match SNI servername"))
+}
+
+func (s *Etcd3Suite) TestCommandStoreConfig(c *check.C) {
+	cmd, display := s.traefikCmd(
+		"storeconfig",
+		withConfigFile("fixtures/simple_web.toml"),
+		"--etcd.endpoint="+ipEtcd+":4001",
+		"--etcd.useAPIV3=true")
+	defer display(c)
+	err := cmd.Start()
+	c.Assert(err, checker.IsNil)
+
+	// wait for traefik finish without error
+	cmd.Wait()
+
+	//CHECK
+	checkmap := map[string]string{
+		"/traefik/loglevel":                 "DEBUG",
+		"/traefik/defaultentrypoints/0":     "http",
+		"/traefik/entrypoints/http/address": ":8000",
+		"/traefik/web/address":              ":8080",
+		"/traefik/etcd/endpoint":            ipEtcd + ":4001",
+	}
+
+	for key, value := range checkmap {
+		var p *store.KVPair
+		err = try.Do(60*time.Second, func() error {
+			p, err = s.kv.Get(key, nil)
+			return err
+		})
+		c.Assert(err, checker.IsNil)
+
+		c.Assert(string(p.Value), checker.Equals, value)
+	}
+}

integration/etcd_test.go

@@ -11,7 +11,7 @@ import (
 	"github.com/containous/traefik/integration/try"
 	"github.com/docker/libkv"
 	"github.com/docker/libkv/store"
-	"github.com/docker/libkv/store/etcd"
+	"github.com/docker/libkv/store/etcd/v2"
 	"github.com/go-check/check"
 
 	checker "github.com/vdemeester/shakers"
@@ -43,7 +43,7 @@ func (s *EtcdSuite) SetUpTest(c *check.C) {
 
 	// wait for etcd
 	err = try.Do(60*time.Second, func() error {
-		_, err := kv.Exists("test")
+		_, err := kv.Exists("test", nil)
 		return err
 	})
 	c.Assert(err, checker.IsNil)
@@ -61,7 +61,13 @@ func (s *EtcdSuite) TearDownSuite(c *check.C) {}
 func (s *EtcdSuite) TestSimpleConfiguration(c *check.C) {
 	etcdHost := s.composeProject.Container(c, "etcd").NetworkSettings.IPAddress
 
-	file := s.adaptFile(c, "fixtures/etcd/simple.toml", struct{ EtcdHost string }{etcdHost})
+	file := s.adaptFile(c, "fixtures/etcd/simple.toml", struct {
+		EtcdHost string
+		UseAPIV3 bool
+	}{
+		etcdHost,
+		false,
+	})
 	defer os.Remove(file)
 
 	cmd, display := s.traefikCmd(withConfigFile(file))
@@ -79,7 +85,13 @@ func (s *EtcdSuite) TestSimpleConfiguration(c *check.C) {
 func (s *EtcdSuite) TestNominalConfiguration(c *check.C) {
 	etcdHost := s.composeProject.Container(c, "etcd").NetworkSettings.IPAddress
 
-	file := s.adaptFile(c, "fixtures/etcd/simple.toml", struct{ EtcdHost string }{etcdHost})
+	file := s.adaptFile(c, "fixtures/etcd/simple.toml", struct {
+		EtcdHost string
+		UseAPIV3 bool
+	}{
+		etcdHost,
+		false,
+	})
 	defer os.Remove(file)
 
 	cmd, display := s.traefikCmd(withConfigFile(file))
@@ -138,7 +150,7 @@ func (s *EtcdSuite) TestNominalConfiguration(c *check.C) {
 
 	// wait for etcd
 	err = try.Do(60*time.Second, func() error {
-		_, err := s.kv.Exists("/traefik/frontends/frontend2/routes/test_2/rule")
+		_, err := s.kv.Exists("/traefik/frontends/frontend2/routes/test_2/rule", nil)
 		return err
 	})
 	c.Assert(err, checker.IsNil)
@@ -196,7 +208,7 @@ func (s *EtcdSuite) TestGlobalConfiguration(c *check.C) {
 
 	// wait for etcd
 	err = try.Do(60*time.Second, func() error {
-		_, err := s.kv.Exists("/traefik/entrypoints/http/address")
+		_, err := s.kv.Exists("/traefik/entrypoints/http/address", nil)
 		return err
 	})
 	c.Assert(err, checker.IsNil)
@@ -261,7 +273,7 @@ func (s *EtcdSuite) TestGlobalConfiguration(c *check.C) {
 
 	// wait for etcd
 	err = try.Do(60*time.Second, func() error {
-		_, err := s.kv.Exists("/traefik/frontends/frontend2/routes/test_2/rule")
+		_, err := s.kv.Exists("/traefik/frontends/frontend2/routes/test_2/rule", nil)
 		return err
 	})
 	c.Assert(err, checker.IsNil)
@@ -414,7 +426,7 @@ func (s *EtcdSuite) TestCommandStoreConfig(c *check.C) {
 	for key, value := range checkmap {
 		var p *store.KVPair
 		err = try.Do(60*time.Second, func() error {
-			p, err = s.kv.Get(key)
+			p, err = s.kv.Get(key, nil)
 			return err
 		})
 		c.Assert(err, checker.IsNil)

integration/fixtures/etcd/simple.toml

@@ -11,6 +11,7 @@ logLevel = "DEBUG"
   endpoint = "{{.EtcdHost}}:2379"
   prefix = "/traefik"
   watch = true
+  useAPIV3 = {{.UseAPIV3}}
 
 [web]
   address = ":8081"

integration/integration_test.go

@@ -43,8 +43,8 @@ func init() {
 		check.Suite(&ConsulSuite{})
 		check.Suite(&DockerSuite{})
 		check.Suite(&DynamoDBSuite{})
-		check.Suite(&ErrorPagesSuite{})
 		check.Suite(&EtcdSuite{})
+		check.Suite(&ErrorPagesSuite{})
 		check.Suite(&EurekaSuite{})
 		check.Suite(&FileSuite{})
 		check.Suite(&GRPCSuite{})
@@ -61,6 +61,7 @@ func init() {
 	if *host {
 		// tests launched from the host
 		check.Suite(&ProxyProtocolSuite{})
+		check.Suite(&Etcd3Suite{})
 	}
 }
 

integration/resources/compose/etcd3.yml

@@ -0,0 +1,62 @@
+version: '2'
+
+services:
+
+  etcd:
+    image: quay.io/coreos/etcd:v3.2.9
+    command: /usr/local/bin/etcd --data-dir=/etcd-data --name node1 --initial-advertise-peer-urls http://172.18.0.2:2380 --listen-peer-urls http://172.18.0.2:2380 --advertise-client-urls http://172.18.0.2:2379,http://172.18.0.2:4001 --listen-client-urls http://172.18.0.2:2379,http://172.18.0.2:4001  --initial-cluster node1=http://172.18.0.2:2380 --debug
+    expose:
+      - 2380
+      - 2379
+      - 4001
+      - 7001
+#    networks:
+#      etcd_net:
+#        ipv4_address: 172.10.1.2
+
+  whoami1:
+    image: emilevauge/whoami
+    # depends_on option activate because libcompose (used by libkermit) does not support fix IP yet...
+    # Remove it ASAP
+    depends_on:
+      - etcd
+#    networks:
+#      etcd_net:
+#        ipv4_address: 172.10.1.3
+
+  whoami2:
+    image: emilevauge/whoami
+    # depends_on option activate because libcompose (used by libkermit) does not support fix IP yet...
+    # Remove it ASAP
+    depends_on:
+      - whoami1
+#    networks:
+#      etcd_net:
+#        ipv4_address: 172.10.1.4
+
+  whoami3:
+    image: emilevauge/whoami
+    # depends_on option activate because libcompose (used by libkermit) does not support fix IP yet...
+    # Remove it ASAP
+    depends_on:
+      - whoami2
+#    networks:
+#      etcd_net:
+#        ipv4_address: 172.10.1.5
+
+  whoami4:
+    image: emilevauge/whoami
+    # depends_on option activate because libcompose (used by libkermit) does not support fix IP yet...
+    # Remove it ASAP
+    depends_on:
+      - whoami3
+#    networks:
+#      etcd_net:
+#        ipv4_address: 172.10.1.6
+
+#networks:
+#  etcd_net:
+#    driver: bridge
+#    ipam:
+#     config:
+#       - subnet: 172.10.1.0/28

integration/try/condition.go

@@ -89,7 +89,7 @@ type DoCondition func() error
 // Verify if a Key exists in the store
 func KVExists(kv store.Store, key string) DoCondition {
 	return func() error {
-		_, err := kv.Exists(key)
+		_, err := kv.Exists(key, nil)
 		return err
 	}
 }

provider/etcd/etcd.go

@@ -3,12 +3,14 @@ package etcd
 import (
 	"fmt"
 
+	"github.com/containous/traefik/log"
 	"github.com/containous/traefik/provider"
 	"github.com/containous/traefik/provider/kv"
 	"github.com/containous/traefik/safe"
 	"github.com/containous/traefik/types"
 	"github.com/docker/libkv/store"
-	"github.com/docker/libkv/store/etcd"
+	"github.com/docker/libkv/store/etcd/v2"
+	"github.com/docker/libkv/store/etcd/v3"
 )
 
 var _ provider.Provider = (*Provider)(nil)
@@ -16,6 +18,7 @@ var _ provider.Provider = (*Provider)(nil)
 // Provider holds configurations of the provider.
 type Provider struct {
 	kv.Provider `mapstructure:",squash" export:"true"`
+	UseAPIV3    bool `description:"Use ETCD API V3" export:"true"`
 }
 
 // Provide allows the etcd provider to Provide configurations to traefik
@@ -31,7 +34,14 @@ func (p *Provider) Provide(configurationChan chan<- types.ConfigMessage, pool *s
 
 // CreateStore creates the KV store
 func (p *Provider) CreateStore() (store.Store, error) {
-	p.SetStoreType(store.ETCD)
+	if p.UseAPIV3 {
-	etcd.Register()
+		etcdv3.Register()
+		p.SetStoreType(store.ETCDV3)
+	} else {
+		// TODO: Deprecated
+		log.Warn("The ETCD API V2 is deprecated. Please use API V3 instead")
+		etcd.Register()
+		p.SetStoreType(store.ETCD)
+	}
 	return p.Provider.CreateStore()
 }

provider/kv/kv.go

@@ -65,7 +65,7 @@ func (p *Provider) SetKVClient(kvClient store.Store) {
 
 func (p *Provider) watchKv(configurationChan chan<- types.ConfigMessage, prefix string, stop chan bool) error {
 	operation := func() error {
-		events, err := p.kvclient.WatchTree(p.Prefix, make(chan struct{}))
+		events, err := p.kvclient.WatchTree(p.Prefix, make(chan struct{}), nil)
 		if err != nil {
 			return fmt.Errorf("Failed to KV WatchTree: %v", err)
 		}
@@ -102,7 +102,7 @@ func (p *Provider) watchKv(configurationChan chan<- types.ConfigMessage, prefix
 func (p *Provider) Provide(configurationChan chan<- types.ConfigMessage, pool *safe.Pool, constraints types.Constraints) error {
 	p.Constraints = append(p.Constraints, constraints...)
 	operation := func() error {
-		if _, err := p.kvclient.Exists(p.Prefix + "/qmslkjdfmqlskdjfmqlksjazçueznbvbwzlkajzebvkwjdcqmlsfj"); err != nil {
+		if _, err := p.kvclient.Exists(p.Prefix+"/qmslkjdfmqlskdjfmqlksjazçueznbvbwzlkajzebvkwjdcqmlsfj", nil); err != nil {
 			return fmt.Errorf("Failed to test KV store connection: %v", err)
 		}
 		if p.Watch {
@@ -165,7 +165,7 @@ func (p *Provider) loadConfig() *types.Configuration {
 
 func (p *Provider) list(keys ...string) []string {
 	joinedKeys := strings.Join(keys, "")
-	keysPairs, err := p.kvclient.List(joinedKeys)
+	keysPairs, err := p.kvclient.List(joinedKeys, nil)
 	if err != nil {
 		log.Debugf("Cannot get keys %s %s ", joinedKeys, err)
 		return nil
@@ -182,7 +182,7 @@ func (p *Provider) listServers(backend string) []string {
 	serverNames := p.list(backend, "/servers/")
 	return fun.Filter(func(serverName string) bool {
 		key := fmt.Sprint(serverName, "/url")
-		if _, err := p.kvclient.Get(key); err != nil {
+		if _, err := p.kvclient.Get(key, nil); err != nil {
 			if err != store.ErrKeyNotFound {
 				log.Errorf("Failed to retrieve value for key %s: %s", key, err)
 			}
@@ -194,7 +194,10 @@ func (p *Provider) listServers(backend string) []string {
 
 func (p *Provider) get(defaultValue string, keys ...string) string {
 	joinedKeys := strings.Join(keys, "")
-	keyPair, err := p.kvclient.Get(strings.TrimPrefix(joinedKeys, "/"))
+	if p.storeType == store.ETCD {
+		joinedKeys = strings.TrimPrefix(joinedKeys, "/")
+	}
+	keyPair, err := p.kvclient.Get(joinedKeys, nil)
 	if err != nil {
 		log.Debugf("Cannot get key %s %s, setting default %s", joinedKeys, err, defaultValue)
 		return defaultValue
@@ -207,7 +210,7 @@ func (p *Provider) get(defaultValue string, keys ...string) string {
 
 func (p *Provider) splitGet(keys ...string) []string {
 	joinedKeys := strings.Join(keys, "")
-	keyPair, err := p.kvclient.Get(joinedKeys)
+	keyPair, err := p.kvclient.Get(joinedKeys, nil)
 	if err != nil {
 		log.Debugf("Cannot get key %s %s, setting default empty", joinedKeys, err)
 		return []string{}
@@ -225,7 +228,7 @@ func (p *Provider) last(key string) string {
 
 func (p *Provider) checkConstraints(keys ...string) bool {
 	joinedKeys := strings.Join(keys, "")
-	keyPair, err := p.kvclient.Get(joinedKeys)
+	keyPair, err := p.kvclient.Get(joinedKeys, nil)
 
 	value := ""
 	if err == nil && keyPair != nil && keyPair.Value != nil {

provider/kv/kv_mock_test.go

@@ -33,7 +33,7 @@ func (s *Mock) Put(key string, value []byte, opts *store.WriteOptions) error {
 	return errors.New("Put not supported")
 }
 
-func (s *Mock) Get(key string) (*store.KVPair, error) {
+func (s *Mock) Get(key string, options *store.ReadOptions) (*store.KVPair, error) {
 	if err := s.Error.Get; err != nil {
 		return nil, err
 	}
@@ -50,7 +50,7 @@ func (s *Mock) Delete(key string) error {
 }
 
 // Exists mock
-func (s *Mock) Exists(key string) (bool, error) {
+func (s *Mock) Exists(key string, options *store.ReadOptions) (bool, error) {
 	if err := s.Error.Get; err != nil {
 		return false, err
 	}
@@ -63,12 +63,12 @@ func (s *Mock) Exists(key string) (bool, error) {
 }
 
 // Watch mock
-func (s *Mock) Watch(key string, stopCh <-chan struct{}) (<-chan *store.KVPair, error) {
+func (s *Mock) Watch(key string, stopCh <-chan struct{}, options *store.ReadOptions) (<-chan *store.KVPair, error) {
 	return nil, errors.New("Watch not supported")
 }
 
 // WatchTree mock
-func (s *Mock) WatchTree(prefix string, stopCh <-chan struct{}) (<-chan []*store.KVPair, error) {
+func (s *Mock) WatchTree(prefix string, stopCh <-chan struct{}, options *store.ReadOptions) (<-chan []*store.KVPair, error) {
 	return s.WatchTreeMethod(), nil
 }
 
@@ -78,7 +78,7 @@ func (s *Mock) NewLock(key string, options *store.LockOptions) (store.Locker, er
 }
 
 // List mock
-func (s *Mock) List(prefix string) ([]*store.KVPair, error) {
+func (s *Mock) List(prefix string, options *store.ReadOptions) ([]*store.KVPair, error) {
 	if err := s.Error.List; err != nil {
 		return nil, err
 	}

provider/kv/kv_test.go

@@ -383,7 +383,7 @@ func TestKVHasStickinessLabel(t *testing.T) {
 			desc: "with cookie name without stickiness=true",
 			KVPairs: []*store.KVPair{
 				{
-					Key:   "loadbalancer/stickiness/cookiename",
+					Key:   "/loadbalancer/stickiness/cookiename",
 					Value: []byte("foo"),
 				},
 			},
@@ -393,7 +393,7 @@ func TestKVHasStickinessLabel(t *testing.T) {
 			desc: "stickiness=true",
 			KVPairs: []*store.KVPair{
 				{
-					Key:   "loadbalancer/stickiness",
+					Key:   "/loadbalancer/stickiness",
 					Value: []byte("true"),
 				},
 			},

vendor/github.com/containous/staert/kv.go

@@ -5,19 +5,20 @@ import (
 	"encoding/base64"
 	"errors"
 	"fmt"
-	"github.com/containous/flaeg"
-	"github.com/docker/libkv"
-	"github.com/docker/libkv/store"
-	"github.com/mitchellh/mapstructure"
 	"reflect"
 	"sort"
 	"strconv"
 	"strings"
+
+	"github.com/containous/flaeg"
+	"github.com/docker/libkv"
+	"github.com/docker/libkv/store"
+	"github.com/mitchellh/mapstructure"
 )
 
 // KvSource implements Source
-// It handles all mapstructure features(Squashed Embeded Sub-Structures, Maps, Pointers)
+// It handles all mapstructure features(Squashed Embedded Sub-Structures, Maps, Pointers)
-// It supports Slices (and maybe Arraies). They must be sorted in the KvStore like this :
+// It supports Slices (and maybe Arrays). They must be sorted in the KvStore like this :
 // Key : ".../[sliceIndex]" -> Value
 type KvSource struct {
 	store.Store
@@ -26,8 +27,8 @@ type KvSource struct {
 
 // NewKvSource creates a new KvSource
 func NewKvSource(backend store.Backend, addrs []string, options *store.Config, prefix string) (*KvSource, error) {
-	store, err := libkv.NewStore(backend, addrs, options)
+	kvStore, err := libkv.NewStore(backend, addrs, options)
-	return &KvSource{Store: store, Prefix: prefix}, err
+	return &KvSource{Store: kvStore, Prefix: prefix}, err
 }
 
 // Parse uses libkv and mapstructure to fill the structure
@@ -46,11 +47,11 @@ func (kv *KvSource) LoadConfig(config interface{}) error {
 		return err
 	}
 	// fmt.Printf("pairs : %#v\n", pairs)
-	mapstruct, err := generateMapstructure(convertPairs(pairs), kv.Prefix)
+	mapStruct, err := generateMapstructure(convertPairs(pairs), kv.Prefix)
 	if err != nil {
 		return err
 	}
-	// fmt.Printf("mapstruct : %#v\n", mapstruct)
+	// fmt.Printf("mapStruct : %#v\n", mapStruct)
 	configDecoder := &mapstructure.DecoderConfig{
 		Metadata:         nil,
 		Result:           config,
@@ -61,7 +62,7 @@ func (kv *KvSource) LoadConfig(config interface{}) error {
 	if err != nil {
 		return err
 	}
-	if err := decoder.Decode(mapstruct); err != nil {
+	if err := decoder.Decode(mapStruct); err != nil {
 		return err
 	}
 	return nil
@@ -82,8 +83,8 @@ func generateMapstructure(pairs []*store.KVPair, prefix string) (map[string]inte
 }
 
 func processKV(key string, v []byte, raw map[string]interface{}) (map[string]interface{}, error) {
-	// Determine which map we're writing the value to. We split by '/'
+	// Determine which map we're writing the value to.
-	// to determine any sub-maps that need to be created.
+	// We split by '/' to determine any sub-maps that need to be created.
 	m := raw
 	children := strings.Split(key, "/")
 	if len(children) > 0 {
@@ -113,7 +114,7 @@ func decodeHook(fromType reflect.Type, toType reflect.Type, data interface{}) (i
 		object := reflect.New(toType.Elem()).Interface()
 		err := object.(encoding.TextUnmarshaler).UnmarshalText([]byte(data.(string)))
 		if err != nil {
-			return nil, fmt.Errorf("Error unmarshaling %v: %v", data, err)
+			return nil, fmt.Errorf("error unmarshaling %v: %v", data, err)
 		}
 		return object, nil
 	}
@@ -170,7 +171,7 @@ func (kv *KvSource) StoreConfig(config interface{}) error {
 	if err := collateKvRecursive(reflect.ValueOf(config), kvMap, kv.Prefix); err != nil {
 		return err
 	}
-	keys := []string{}
+	var keys []string
 	for key := range kvMap {
 		keys = append(keys, key)
 	}
@@ -198,7 +199,7 @@ func collateKvRecursive(objValue reflect.Value, kv map[string]string, key string
 	if marshaler, ok := objValue.Interface().(encoding.TextMarshaler); ok {
 		test, err := marshaler.MarshalText()
 		if err != nil {
-			return fmt.Errorf("Error marshaling key %s: %v", name, err)
+			return fmt.Errorf("error marshaling key %s: %v", name, err)
 		}
 		kv[name] = string(test)
 		return nil
@@ -252,7 +253,7 @@ func collateKvRecursive(objValue reflect.Value, kv map[string]string, key string
 	case reflect.Map:
 		for _, k := range objValue.MapKeys() {
 			if k.Kind() == reflect.Struct {
-				return errors.New("Struct as key not supported")
+				return errors.New("struct as key not supported")
 			}
 			name = key + "/" + fmt.Sprint(k)
 			if err := collateKvRecursive(objValue.MapIndex(k), kv, name); err != nil {
@@ -280,14 +281,14 @@ func collateKvRecursive(objValue reflect.Value, kv map[string]string, key string
 		kv[name] = fmt.Sprint(objValue)
 
 	default:
-		return fmt.Errorf("Kind %s not supported", kind.String())
+		return fmt.Errorf("kind %s not supported", kind.String())
 	}
 	return nil
 }
 
-// ListRecursive lists all key value childrens under key
+// ListRecursive lists all key value children under key
 func (kv *KvSource) ListRecursive(key string, pairs map[string][]byte) error {
-	pairsN1, err := kv.List(key)
+	pairsN1, err := kv.List(key, nil)
 	if err == store.ErrKeyNotFound {
 		return nil
 	}
@@ -295,7 +296,7 @@ func (kv *KvSource) ListRecursive(key string, pairs map[string][]byte) error {
 		return err
 	}
 	if len(pairsN1) == 0 {
-		pairLeaf, err := kv.Get(key)
+		pairLeaf, err := kv.Get(key, nil)
 		if err != nil {
 			return err
 		}

vendor/github.com/containous/staert/staert.go

@@ -2,12 +2,13 @@ package staert
 
 import (
 	"fmt"
-	"github.com/BurntSushi/toml"
-	"github.com/containous/flaeg"
 	"os"
 	"path/filepath"
 	"reflect"
 	"strings"
+
+	"github.com/BurntSushi/toml"
+	"github.com/containous/flaeg"
 )
 
 // Source interface must be satisfy to Add any kink of Source to Staert as like as TomlFile or Flaeg
@@ -21,7 +22,7 @@ type Staert struct {
 	sources []Source
 }
 
-// NewStaert creats and return a pointer on Staert. Need defaultConfig and defaultPointersConfig given by references
+// NewStaert creates and return a pointer on Staert. Need defaultConfig and defaultPointersConfig given by references
 func NewStaert(rootCommand *flaeg.Command) *Staert {
 	s := Staert{
 		command: rootCommand,
@@ -62,7 +63,7 @@ func (s *Staert) LoadConfig() (interface{}, error) {
 					fCmdConfigType := reflect.TypeOf(fCmd.Config)
 					sCmdConfigType := reflect.TypeOf(s.command.Config)
 					if fCmdConfigType != sCmdConfigType {
-						return nil, fmt.Errorf("Command %s : Config type doesn't match with root command config type. Expected %s got %s", fCmd.Name, sCmdConfigType.Name(), fCmdConfigType.Name())
+						return nil, fmt.Errorf("command %s : Config type doesn't match with root command config type. Expected %s got %s", fCmd.Name, sCmdConfigType.Name(), fCmdConfigType.Name())
 					}
 					s.command = fCmd
 				} else {
@@ -90,7 +91,7 @@ type TomlSource struct {
 	fullpath     string
 }
 
-// NewTomlSource creats and return a pointer on TomlSource.
+// NewTomlSource creates and return a pointer on TomlSource.
 // Parameter filename is the file name (without extension type, ".toml" will be added)
 // dirNfullpath may contain directories or fullpath to the file.
 func NewTomlSource(filename string, dirNfullpath []string) *TomlSource {
@@ -118,13 +119,13 @@ func preprocessDir(dirIn string) (string, error) {
 func findFile(filename string, dirNfile []string) string {
 	for _, df := range dirNfile {
 		if df != "" {
-			fullpath, _ := preprocessDir(df)
+			fullPath, _ := preprocessDir(df)
-			if fileinfo, err := os.Stat(fullpath); err == nil && !fileinfo.IsDir() {
+			if fileInfo, err := os.Stat(fullPath); err == nil && !fileInfo.IsDir() {
-				return fullpath
+				return fullPath
 			}
-			fullpath = fullpath + "/" + filename + ".toml"
+			fullPath = fullPath + "/" + filename + ".toml"
-			if fileinfo, err := os.Stat(fullpath); err == nil && !fileinfo.IsDir() {
+			if fileInfo, err := os.Stat(fullPath); err == nil && !fileInfo.IsDir() {
-				return fullpath
+				return fullPath
 			}
 		}
 	}
@@ -167,7 +168,7 @@ func (ts *TomlSource) Parse(cmd *flaeg.Command) (*flaeg.Command, error) {
 }
 
 func generateArgs(metadata toml.MetaData, flags []string) ([]string, bool, error) {
-	flaegArgs := []string{}
+	var flaegArgs []string
 	keys := metadata.Keys()
 	hasUnderField := false
 	for i, key := range keys {

vendor/github.com/coreos/bbolt/LICENSE

@@ -0,0 +1,20 @@
+The MIT License (MIT)
+
+Copyright (c) 2013 Ben Johnson
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+the Software, and to permit persons to whom the Software is furnished to do so,
+subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

vendor/github.com/coreos/bbolt/bolt_386.go

@@ -0,0 +1,10 @@
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0x7FFFFFFF // 2GB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0xFFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_amd64.go

@@ -0,0 +1,10 @@
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0xFFFFFFFFFFFF // 256TB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0x7FFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_arm.go

@@ -0,0 +1,28 @@
+package bolt
+
+import "unsafe"
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0x7FFFFFFF // 2GB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0xFFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned bool
+
+func init() {
+	// Simple check to see whether this arch handles unaligned load/stores
+	// correctly.
+
+	// ARM9 and older devices require load/stores to be from/to aligned
+	// addresses. If not, the lower 2 bits are cleared and that address is
+	// read in a jumbled up order.
+
+	// See http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.faqs/ka15414.html
+
+	raw := [6]byte{0xfe, 0xef, 0x11, 0x22, 0x22, 0x11}
+	val := *(*uint32)(unsafe.Pointer(uintptr(unsafe.Pointer(&raw)) + 2))
+
+	brokenUnaligned = val != 0x11222211
+}

vendor/github.com/coreos/bbolt/bolt_arm64.go

@@ -0,0 +1,12 @@
+// +build arm64
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0xFFFFFFFFFFFF // 256TB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0x7FFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_linux.go

@@ -0,0 +1,10 @@
+package bolt
+
+import (
+	"syscall"
+)
+
+// fdatasync flushes written data to a file descriptor.
+func fdatasync(db *DB) error {
+	return syscall.Fdatasync(int(db.file.Fd()))
+}

vendor/github.com/coreos/bbolt/bolt_mips64x.go

@@ -0,0 +1,12 @@
+// +build mips64 mips64le
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0x8000000000 // 512GB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0x7FFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_mipsx.go

@@ -0,0 +1,12 @@
+// +build mips mipsle
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0x40000000 // 1GB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0xFFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_openbsd.go

@@ -0,0 +1,27 @@
+package bolt
+
+import (
+	"syscall"
+	"unsafe"
+)
+
+const (
+	msAsync      = 1 << iota // perform asynchronous writes
+	msSync                   // perform synchronous writes
+	msInvalidate             // invalidate cached data
+)
+
+func msync(db *DB) error {
+	_, _, errno := syscall.Syscall(syscall.SYS_MSYNC, uintptr(unsafe.Pointer(db.data)), uintptr(db.datasz), msInvalidate)
+	if errno != 0 {
+		return errno
+	}
+	return nil
+}
+
+func fdatasync(db *DB) error {
+	if db.data != nil {
+		return msync(db)
+	}
+	return db.file.Sync()
+}

vendor/github.com/coreos/bbolt/bolt_ppc.go

@@ -0,0 +1,9 @@
+// +build ppc
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0x7FFFFFFF // 2GB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0xFFFFFFF

vendor/github.com/coreos/bbolt/bolt_ppc64.go

@@ -0,0 +1,12 @@
+// +build ppc64
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0xFFFFFFFFFFFF // 256TB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0x7FFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_ppc64le.go

@@ -0,0 +1,12 @@
+// +build ppc64le
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0xFFFFFFFFFFFF // 256TB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0x7FFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_s390x.go

@@ -0,0 +1,12 @@
+// +build s390x
+
+package bolt
+
+// maxMapSize represents the largest mmap size supported by Bolt.
+const maxMapSize = 0xFFFFFFFFFFFF // 256TB
+
+// maxAllocSize is the size used when creating array pointers.
+const maxAllocSize = 0x7FFFFFFF
+
+// Are unaligned load/stores broken on this arch?
+var brokenUnaligned = false

vendor/github.com/coreos/bbolt/bolt_unix.go

@@ -0,0 +1,92 @@
+// +build !windows,!plan9,!solaris
+
+package bolt
+
+import (
+	"fmt"
+	"os"
+	"syscall"
+	"time"
+	"unsafe"
+)
+
+// flock acquires an advisory lock on a file descriptor.
+func flock(db *DB, mode os.FileMode, exclusive bool, timeout time.Duration) error {
+	var t time.Time
+	if timeout != 0 {
+		t = time.Now()
+	}
+	fd := db.file.Fd()
+	flag := syscall.LOCK_NB
+	if exclusive {
+		flag |= syscall.LOCK_EX
+	} else {
+		flag |= syscall.LOCK_SH
+	}
+	for {
+		// Attempt to obtain an exclusive lock.
+		err := syscall.Flock(int(fd), flag)
+		if err == nil {
+			return nil
+		} else if err != syscall.EWOULDBLOCK {
+			return err
+		}
+
+		// If we timed out then return an error.
+		if timeout != 0 && time.Since(t) > timeout-flockRetryTimeout {
+			return ErrTimeout
+		}
+
+		// Wait for a bit and try again.
+		time.Sleep(flockRetryTimeout)
+	}
+}
+
+// funlock releases an advisory lock on a file descriptor.
+func funlock(db *DB) error {
+	return syscall.Flock(int(db.file.Fd()), syscall.LOCK_UN)
+}
+
+// mmap memory maps a DB's data file.
+func mmap(db *DB, sz int) error {
+	// Map the data file to memory.
+	b, err := syscall.Mmap(int(db.file.Fd()), 0, sz, syscall.PROT_READ, syscall.MAP_SHARED|db.MmapFlags)
+	if err != nil {
+		return err
+	}
+
+	// Advise the kernel that the mmap is accessed randomly.
+	if err := madvise(b, syscall.MADV_RANDOM); err != nil {
+		return fmt.Errorf("madvise: %s", err)
+	}
+
+	// Save the original byte slice and convert to a byte array pointer.
+	db.dataref = b
+	db.data = (*[maxMapSize]byte)(unsafe.Pointer(&b[0]))
+	db.datasz = sz
+	return nil
+}
+
+// munmap unmaps a DB's data file from memory.
+func munmap(db *DB) error {
+	// Ignore the unmap if we have no mapped data.
+	if db.dataref == nil {
+		return nil
+	}
+
+	// Unmap using the original byte slice.
+	err := syscall.Munmap(db.dataref)
+	db.dataref = nil
+	db.data = nil
+	db.datasz = 0
+	return err
+}
+
+// NOTE: This function is copied from stdlib because it is not available on darwin.
+func madvise(b []byte, advice int) (err error) {
+	_, _, e1 := syscall.Syscall(syscall.SYS_MADVISE, uintptr(unsafe.Pointer(&b[0])), uintptr(len(b)), uintptr(advice))
+	if e1 != 0 {
+		err = e1
+	}
+	return
+}

vendor/github.com/coreos/bbolt/bolt_unix_solaris.go

@@ -0,0 +1,89 @@
+package bolt
+
+import (
+	"fmt"
+	"os"
+	"syscall"
+	"time"
+	"unsafe"
+
+	"golang.org/x/sys/unix"
+)
+
+// flock acquires an advisory lock on a file descriptor.
+func flock(db *DB, mode os.FileMode, exclusive bool, timeout time.Duration) error {
+	var t time.Time
+	if timeout != 0 {
+		t = time.Now()
+	}
+	fd := db.file.Fd()
+	var lockType int16
+	if exclusive {
+		lockType = syscall.F_WRLCK
+	} else {
+		lockType = syscall.F_RDLCK
+	}
+	for {
+		// Attempt to obtain an exclusive lock.
+		lock := syscall.Flock_t{Type: lockType}
+		err := syscall.FcntlFlock(fd, syscall.F_SETLK, &lock)
+		if err == nil {
+			return nil
+		} else if err != syscall.EAGAIN {
+			return err
+		}
+
+		// If we timed out then return an error.
+		if timeout != 0 && time.Since(t) > timeout-flockRetryTimeout {
+			return ErrTimeout
+		}
+
+		// Wait for a bit and try again.
+		time.Sleep(flockRetryTimeout)
+	}
+}
+
+// funlock releases an advisory lock on a file descriptor.
+func funlock(db *DB) error {
+	var lock syscall.Flock_t
+	lock.Start = 0
+	lock.Len = 0
+	lock.Type = syscall.F_UNLCK
+	lock.Whence = 0
+	return syscall.FcntlFlock(uintptr(db.file.Fd()), syscall.F_SETLK, &lock)
+}
+
+// mmap memory maps a DB's data file.
+func mmap(db *DB, sz int) error {
+	// Map the data file to memory.
+	b, err := unix.Mmap(int(db.file.Fd()), 0, sz, syscall.PROT_READ, syscall.MAP_SHARED|db.MmapFlags)
+	if err != nil {
+		return err
+	}
+
+	// Advise the kernel that the mmap is accessed randomly.
+	if err := unix.Madvise(b, syscall.MADV_RANDOM); err != nil {
+		return fmt.Errorf("madvise: %s", err)
+	}
+
+	// Save the original byte slice and convert to a byte array pointer.
+	db.dataref = b
+	db.data = (*[maxMapSize]byte)(unsafe.Pointer(&b[0]))
+	db.datasz = sz
+	return nil
+}
+
+// munmap unmaps a DB's data file from memory.
+func munmap(db *DB) error {
+	// Ignore the unmap if we have no mapped data.
+	if db.dataref == nil {
+		return nil
+	}
+
+	// Unmap using the original byte slice.
+	err := unix.Munmap(db.dataref)
+	db.dataref = nil
+	db.data = nil
+	db.datasz = 0
+	return err
+}

vendor/github.com/coreos/bbolt/bolt_windows.go

@@ -0,0 +1,145 @@
+package bolt
+
+import (
+	"fmt"
+	"os"
+	"syscall"
+	"time"
+	"unsafe"
+)
+
+// LockFileEx code derived from golang build filemutex_windows.go @ v1.5.1
+var (
+	modkernel32      = syscall.NewLazyDLL("kernel32.dll")
+	procLockFileEx   = modkernel32.NewProc("LockFileEx")
+	procUnlockFileEx = modkernel32.NewProc("UnlockFileEx")
+)
+
+const (
+	lockExt = ".lock"
+
+	// see https://msdn.microsoft.com/en-us/library/windows/desktop/aa365203(v=vs.85).aspx
+	flagLockExclusive       = 2
+	flagLockFailImmediately = 1
+
+	// see https://msdn.microsoft.com/en-us/library/windows/desktop/ms681382(v=vs.85).aspx
+	errLockViolation syscall.Errno = 0x21
+)
+
+func lockFileEx(h syscall.Handle, flags, reserved, locklow, lockhigh uint32, ol *syscall.Overlapped) (err error) {
+	r, _, err := procLockFileEx.Call(uintptr(h), uintptr(flags), uintptr(reserved), uintptr(locklow), uintptr(lockhigh), uintptr(unsafe.Pointer(ol)))
+	if r == 0 {
+		return err
+	}
+	return nil
+}
+
+func unlockFileEx(h syscall.Handle, reserved, locklow, lockhigh uint32, ol *syscall.Overlapped) (err error) {
+	r, _, err := procUnlockFileEx.Call(uintptr(h), uintptr(reserved), uintptr(locklow), uintptr(lockhigh), uintptr(unsafe.Pointer(ol)), 0)
+	if r == 0 {
+		return err
+	}
+	return nil
+}
+
+// fdatasync flushes written data to a file descriptor.
+func fdatasync(db *DB) error {
+	return db.file.Sync()
+}
+
+// flock acquires an advisory lock on a file descriptor.
+func flock(db *DB, mode os.FileMode, exclusive bool, timeout time.Duration) error {
+	// Create a separate lock file on windows because a process
+	// cannot share an exclusive lock on the same file. This is
+	// needed during Tx.WriteTo().
+	f, err := os.OpenFile(db.path+lockExt, os.O_CREATE, mode)
+	if err != nil {
+		return err
+	}
+	db.lockfile = f
+
+	var t time.Time
+	if timeout != 0 {
+		t = time.Now()
+	}
+	fd := f.Fd()
+	var flag uint32 = flagLockFailImmediately
+	if exclusive {
+		flag |= flagLockExclusive
+	}
+	for {
+		// Attempt to obtain an exclusive lock.
+		err := lockFileEx(syscall.Handle(fd), flag, 0, 1, 0, &syscall.Overlapped{})
+		if err == nil {
+			return nil
+		} else if err != errLockViolation {
+			return err
+		}
+
+		// If we timed oumercit then return an error.
+		if timeout != 0 && time.Since(t) > timeout-flockRetryTimeout {
+			return ErrTimeout
+		}
+
+		// Wait for a bit and try again.
+		time.Sleep(flockRetryTimeout)
+	}
+}
+
+// funlock releases an advisory lock on a file descriptor.
+func funlock(db *DB) error {
+	err := unlockFileEx(syscall.Handle(db.lockfile.Fd()), 0, 1, 0, &syscall.Overlapped{})
+	db.lockfile.Close()
+	os.Remove(db.path + lockExt)
+	return err
+}
+
+// mmap memory maps a DB's data file.
+// Based on: https://github.com/edsrzf/mmap-go
+func mmap(db *DB, sz int) error {
+	if !db.readOnly {
+		// Truncate the database to the size of the mmap.
+		if err := db.file.Truncate(int64(sz)); err != nil {
+			return fmt.Errorf("truncate: %s", err)
+		}
+	}
+
+	// Open a file mapping handle.
+	sizelo := uint32(sz >> 32)
+	sizehi := uint32(sz) & 0xffffffff
+	h, errno := syscall.CreateFileMapping(syscall.Handle(db.file.Fd()), nil, syscall.PAGE_READONLY, sizelo, sizehi, nil)
+	if h == 0 {
+		return os.NewSyscallError("CreateFileMapping", errno)
+	}
+
+	// Create the memory map.
+	addr, errno := syscall.MapViewOfFile(h, syscall.FILE_MAP_READ, 0, 0, uintptr(sz))
+	if addr == 0 {
+		return os.NewSyscallError("MapViewOfFile", errno)
+	}
+
+	// Close mapping handle.
+	if err := syscall.CloseHandle(syscall.Handle(h)); err != nil {
+		return os.NewSyscallError("CloseHandle", err)
+	}
+
+	// Convert to a byte array.
+	db.data = ((*[maxMapSize]byte)(unsafe.Pointer(addr)))
+	db.datasz = sz
+
+	return nil
+}
+
+// munmap unmaps a pointer from a file.
+// Based on: https://github.com/edsrzf/mmap-go
+func munmap(db *DB) error {
+	if db.data == nil {
+		return nil
+	}
+
+	addr := (uintptr)(unsafe.Pointer(&db.data[0]))
+	if err := syscall.UnmapViewOfFile(addr); err != nil {
+		return os.NewSyscallError("UnmapViewOfFile", err)
+	}
+	return nil
+}

vendor/github.com/coreos/bbolt/boltsync_unix.go

@@ -0,0 +1,8 @@
+// +build !windows,!plan9,!linux,!openbsd
+
+package bolt
+
+// fdatasync flushes written data to a file descriptor.
+func fdatasync(db *DB) error {
+	return db.file.Sync()
+}

vendor/github.com/coreos/bbolt/bucket.go

@@ -0,0 +1,775 @@
+package bolt
+
+import (
+	"bytes"
+	"fmt"
+	"unsafe"
+)
+
+const (
+	// MaxKeySize is the maximum length of a key, in bytes.
+	MaxKeySize = 32768
+
+	// MaxValueSize is the maximum length of a value, in bytes.
+	MaxValueSize = (1 << 31) - 2
+)
+
+const bucketHeaderSize = int(unsafe.Sizeof(bucket{}))
+
+const (
+	minFillPercent = 0.1
+	maxFillPercent = 1.0
+)
+
+// DefaultFillPercent is the percentage that split pages are filled.
+// This value can be changed by setting Bucket.FillPercent.
+const DefaultFillPercent = 0.5
+
+// Bucket represents a collection of key/value pairs inside the database.
+type Bucket struct {
+	*bucket
+	tx       *Tx                // the associated transaction
+	buckets  map[string]*Bucket // subbucket cache
+	page     *page              // inline page reference
+	rootNode *node              // materialized node for the root page.
+	nodes    map[pgid]*node     // node cache
+
+	// Sets the threshold for filling nodes when they split. By default,
+	// the bucket will fill to 50% but it can be useful to increase this
+	// amount if you know that your write workloads are mostly append-only.
+	//
+	// This is non-persisted across transactions so it must be set in every Tx.
+	FillPercent float64
+}
+
+// bucket represents the on-file representation of a bucket.
+// This is stored as the "value" of a bucket key. If the bucket is small enough,
+// then its root page can be stored inline in the "value", after the bucket
+// header. In the case of inline buckets, the "root" will be 0.
+type bucket struct {
+	root     pgid   // page id of the bucket's root-level page
+	sequence uint64 // monotonically incrementing, used by NextSequence()
+}
+
+// newBucket returns a new bucket associated with a transaction.
+func newBucket(tx *Tx) Bucket {
+	var b = Bucket{tx: tx, FillPercent: DefaultFillPercent}
+	if tx.writable {
+		b.buckets = make(map[string]*Bucket)
+		b.nodes = make(map[pgid]*node)
+	}
+	return b
+}
+
+// Tx returns the tx of the bucket.
+func (b *Bucket) Tx() *Tx {
+	return b.tx
+}
+
+// Root returns the root of the bucket.
+func (b *Bucket) Root() pgid {
+	return b.root
+}
+
+// Writable returns whether the bucket is writable.
+func (b *Bucket) Writable() bool {
+	return b.tx.writable
+}
+
+// Cursor creates a cursor associated with the bucket.
+// The cursor is only valid as long as the transaction is open.
+// Do not use a cursor after the transaction is closed.
+func (b *Bucket) Cursor() *Cursor {
+	// Update transaction statistics.
+	b.tx.stats.CursorCount++
+
+	// Allocate and return a cursor.
+	return &Cursor{
+		bucket: b,
+		stack:  make([]elemRef, 0),
+	}
+}
+
+// Bucket retrieves a nested bucket by name.
+// Returns nil if the bucket does not exist.
+// The bucket instance is only valid for the lifetime of the transaction.
+func (b *Bucket) Bucket(name []byte) *Bucket {
+	if b.buckets != nil {
+		if child := b.buckets[string(name)]; child != nil {
+			return child
+		}
+	}
+
+	// Move cursor to key.
+	c := b.Cursor()
+	k, v, flags := c.seek(name)
+
+	// Return nil if the key doesn't exist or it is not a bucket.
+	if !bytes.Equal(name, k) || (flags&bucketLeafFlag) == 0 {
+		return nil
+	}
+
+	// Otherwise create a bucket and cache it.
+	var child = b.openBucket(v)
+	if b.buckets != nil {
+		b.buckets[string(name)] = child
+	}
+
+	return child
+}
+
+// Helper method that re-interprets a sub-bucket value
+// from a parent into a Bucket
+func (b *Bucket) openBucket(value []byte) *Bucket {
+	var child = newBucket(b.tx)
+
+	// If unaligned load/stores are broken on this arch and value is
+	// unaligned simply clone to an aligned byte array.
+	unaligned := brokenUnaligned && uintptr(unsafe.Pointer(&value[0]))&3 != 0
+
+	if unaligned {
+		value = cloneBytes(value)
+	}
+
+	// If this is a writable transaction then we need to copy the bucket entry.
+	// Read-only transactions can point directly at the mmap entry.
+	if b.tx.writable && !unaligned {
+		child.bucket = &bucket{}
+		*child.bucket = *(*bucket)(unsafe.Pointer(&value[0]))
+	} else {
+		child.bucket = (*bucket)(unsafe.Pointer(&value[0]))
+	}
+
+	// Save a reference to the inline page if the bucket is inline.
+	if child.root == 0 {
+		child.page = (*page)(unsafe.Pointer(&value[bucketHeaderSize]))
+	}
+
+	return &child
+}
+
+// CreateBucket creates a new bucket at the given key and returns the new bucket.
+// Returns an error if the key already exists, if the bucket name is blank, or if the bucket name is too long.
+// The bucket instance is only valid for the lifetime of the transaction.
+func (b *Bucket) CreateBucket(key []byte) (*Bucket, error) {
+	if b.tx.db == nil {
+		return nil, ErrTxClosed
+	} else if !b.tx.writable {
+		return nil, ErrTxNotWritable
+	} else if len(key) == 0 {
+		return nil, ErrBucketNameRequired
+	}
+
+	// Move cursor to correct position.
+	c := b.Cursor()
+	k, _, flags := c.seek(key)
+
+	// Return an error if there is an existing key.
+	if bytes.Equal(key, k) {
+		if (flags & bucketLeafFlag) != 0 {
+			return nil, ErrBucketExists
+		}
+		return nil, ErrIncompatibleValue
+	}
+
+	// Create empty, inline bucket.
+	var bucket = Bucket{
+		bucket:      &bucket{},
+		rootNode:    &node{isLeaf: true},
+		FillPercent: DefaultFillPercent,
+	}
+	var value = bucket.write()
+
+	// Insert into node.
+	key = cloneBytes(key)
+	c.node().put(key, key, value, 0, bucketLeafFlag)
+
+	// Since subbuckets are not allowed on inline buckets, we need to
+	// dereference the inline page, if it exists. This will cause the bucket
+	// to be treated as a regular, non-inline bucket for the rest of the tx.
+	b.page = nil
+
+	return b.Bucket(key), nil
+}
+
+// CreateBucketIfNotExists creates a new bucket if it doesn't already exist and returns a reference to it.
+// Returns an error if the bucket name is blank, or if the bucket name is too long.
+// The bucket instance is only valid for the lifetime of the transaction.
+func (b *Bucket) CreateBucketIfNotExists(key []byte) (*Bucket, error) {
+	child, err := b.CreateBucket(key)
+	if err == ErrBucketExists {
+		return b.Bucket(key), nil
+	} else if err != nil {
+		return nil, err
+	}
+	return child, nil
+}
+
+// DeleteBucket deletes a bucket at the given key.
+// Returns an error if the bucket does not exists, or if the key represents a non-bucket value.
+func (b *Bucket) DeleteBucket(key []byte) error {
+	if b.tx.db == nil {
+		return ErrTxClosed
+	} else if !b.Writable() {
+		return ErrTxNotWritable
+	}
+
+	// Move cursor to correct position.
+	c := b.Cursor()
+	k, _, flags := c.seek(key)
+
+	// Return an error if bucket doesn't exist or is not a bucket.
+	if !bytes.Equal(key, k) {
+		return ErrBucketNotFound
+	} else if (flags & bucketLeafFlag) == 0 {
+		return ErrIncompatibleValue
+	}
+
+	// Recursively delete all child buckets.
+	child := b.Bucket(key)
+	err := child.ForEach(func(k, v []byte) error {
+		if v == nil {
+			if err := child.DeleteBucket(k); err != nil {
+				return fmt.Errorf("delete bucket: %s", err)
+			}
+		}
+		return nil
+	})
+	if err != nil {
+		return err
+	}
+
+	// Remove cached copy.
+	delete(b.buckets, string(key))
+
+	// Release all bucket pages to freelist.
+	child.nodes = nil
+	child.rootNode = nil
+	child.free()
+
+	// Delete the node if we have a matching key.
+	c.node().del(key)
+
+	return nil
+}
+
+// Get retrieves the value for a key in the bucket.
+// Returns a nil value if the key does not exist or if the key is a nested bucket.
+// The returned value is only valid for the life of the transaction.
+func (b *Bucket) Get(key []byte) []byte {
+	k, v, flags := b.Cursor().seek(key)
+
+	// Return nil if this is a bucket.
+	if (flags & bucketLeafFlag) != 0 {
+		return nil
+	}
+
+	// If our target node isn't the same key as what's passed in then return nil.
+	if !bytes.Equal(key, k) {
+		return nil
+	}
+	return v
+}
+
+// Put sets the value for a key in the bucket.
+// If the key exist then its previous value will be overwritten.
+// Supplied value must remain valid for the life of the transaction.
+// Returns an error if the bucket was created from a read-only transaction, if the key is blank, if the key is too large, or if the value is too large.
+func (b *Bucket) Put(key []byte, value []byte) error {
+	if b.tx.db == nil {
+		return ErrTxClosed
+	} else if !b.Writable() {
+		return ErrTxNotWritable
+	} else if len(key) == 0 {
+		return ErrKeyRequired
+	} else if len(key) > MaxKeySize {
+		return ErrKeyTooLarge
+	} else if int64(len(value)) > MaxValueSize {
+		return ErrValueTooLarge
+	}
+
+	// Move cursor to correct position.
+	c := b.Cursor()
+	k, _, flags := c.seek(key)
+
+	// Return an error if there is an existing key with a bucket value.
+	if bytes.Equal(key, k) && (flags&bucketLeafFlag) != 0 {
+		return ErrIncompatibleValue
+	}
+
+	// Insert into node.
+	key = cloneBytes(key)
+	c.node().put(key, key, value, 0, 0)
+
+	return nil
+}
+
+// Delete removes a key from the bucket.
+// If the key does not exist then nothing is done and a nil error is returned.
+// Returns an error if the bucket was created from a read-only transaction.
+func (b *Bucket) Delete(key []byte) error {
+	if b.tx.db == nil {
+		return ErrTxClosed
+	} else if !b.Writable() {
+		return ErrTxNotWritable
+	}
+
+	// Move cursor to correct position.
+	c := b.Cursor()
+	k, _, flags := c.seek(key)
+
+	// Return nil if the key doesn't exist.
+	if !bytes.Equal(key, k) {
+		return nil
+	}
+
+	// Return an error if there is already existing bucket value.
+	if (flags & bucketLeafFlag) != 0 {
+		return ErrIncompatibleValue
+	}
+
+	// Delete the node if we have a matching key.
+	c.node().del(key)
+
+	return nil
+}
+
+// Sequence returns the current integer for the bucket without incrementing it.
+func (b *Bucket) Sequence() uint64 { return b.bucket.sequence }
+
+// SetSequence updates the sequence number for the bucket.
+func (b *Bucket) SetSequence(v uint64) error {
+	if b.tx.db == nil {
+		return ErrTxClosed
+	} else if !b.Writable() {
+		return ErrTxNotWritable
+	}
+
+	// Materialize the root node if it hasn't been already so that the
+	// bucket will be saved during commit.
+	if b.rootNode == nil {
+		_ = b.node(b.root, nil)
+	}
+
+	// Increment and return the sequence.
+	b.bucket.sequence = v
+	return nil
+}
+
+// NextSequence returns an autoincrementing integer for the bucket.
+func (b *Bucket) NextSequence() (uint64, error) {
+	if b.tx.db == nil {
+		return 0, ErrTxClosed
+	} else if !b.Writable() {
+		return 0, ErrTxNotWritable
+	}
+
+	// Materialize the root node if it hasn't been already so that the
+	// bucket will be saved during commit.
+	if b.rootNode == nil {
+		_ = b.node(b.root, nil)
+	}
+
+	// Increment and return the sequence.
+	b.bucket.sequence++
+	return b.bucket.sequence, nil
+}
+
+// ForEach executes a function for each key/value pair in a bucket.
+// If the provided function returns an error then the iteration is stopped and
+// the error is returned to the caller. The provided function must not modify
+// the bucket; this will result in undefined behavior.
+func (b *Bucket) ForEach(fn func(k, v []byte) error) error {
+	if b.tx.db == nil {
+		return ErrTxClosed
+	}
+	c := b.Cursor()
+	for k, v := c.First(); k != nil; k, v = c.Next() {
+		if err := fn(k, v); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// Stat returns stats on a bucket.
+func (b *Bucket) Stats() BucketStats {
+	var s, subStats BucketStats
+	pageSize := b.tx.db.pageSize
+	s.BucketN += 1
+	if b.root == 0 {
+		s.InlineBucketN += 1
+	}
+	b.forEachPage(func(p *page, depth int) {
+		if (p.flags & leafPageFlag) != 0 {
+			s.KeyN += int(p.count)
+
+			// used totals the used bytes for the page
+			used := pageHeaderSize
+
+			if p.count != 0 {
+				// If page has any elements, add all element headers.
+				used += leafPageElementSize * int(p.count-1)
+
+				// Add all element key, value sizes.
+				// The computation takes advantage of the fact that the position
+				// of the last element's key/value equals to the total of the sizes
+				// of all previous elements' keys and values.
+				// It also includes the last element's header.
+				lastElement := p.leafPageElement(p.count - 1)
+				used += int(lastElement.pos + lastElement.ksize + lastElement.vsize)
+			}
+
+			if b.root == 0 {
+				// For inlined bucket just update the inline stats
+				s.InlineBucketInuse += used
+			} else {
+				// For non-inlined bucket update all the leaf stats
+				s.LeafPageN++
+				s.LeafInuse += used
+				s.LeafOverflowN += int(p.overflow)
+
+				// Collect stats from sub-buckets.
+				// Do that by iterating over all element headers
+				// looking for the ones with the bucketLeafFlag.
+				for i := uint16(0); i < p.count; i++ {
+					e := p.leafPageElement(i)
+					if (e.flags & bucketLeafFlag) != 0 {
+						// For any bucket element, open the element value
+						// and recursively call Stats on the contained bucket.
+						subStats.Add(b.openBucket(e.value()).Stats())
+					}
+				}
+			}
+		} else if (p.flags & branchPageFlag) != 0 {
+			s.BranchPageN++
+			lastElement := p.branchPageElement(p.count - 1)
+
+			// used totals the used bytes for the page
+			// Add header and all element headers.
+			used := pageHeaderSize + (branchPageElementSize * int(p.count-1))
+
+			// Add size of all keys and values.
+			// Again, use the fact that last element's position equals to
+			// the total of key, value sizes of all previous elements.
+			used += int(lastElement.pos + lastElement.ksize)
+			s.BranchInuse += used
+			s.BranchOverflowN += int(p.overflow)
+		}
+
+		// Keep track of maximum page depth.
+		if depth+1 > s.Depth {
+			s.Depth = (depth + 1)
+		}
+	})
+
+	// Alloc stats can be computed from page counts and pageSize.
+	s.BranchAlloc = (s.BranchPageN + s.BranchOverflowN) * pageSize
+	s.LeafAlloc = (s.LeafPageN + s.LeafOverflowN) * pageSize
+
+	// Add the max depth of sub-buckets to get total nested depth.
+	s.Depth += subStats.Depth
+	// Add the stats for all sub-buckets
+	s.Add(subStats)
+	return s
+}
+
+// forEachPage iterates over every page in a bucket, including inline pages.
+func (b *Bucket) forEachPage(fn func(*page, int)) {
+	// If we have an inline page then just use that.
+	if b.page != nil {
+		fn(b.page, 0)
+		return
+	}
+
+	// Otherwise traverse the page hierarchy.
+	b.tx.forEachPage(b.root, 0, fn)
+}
+
+// forEachPageNode iterates over every page (or node) in a bucket.
+// This also includes inline pages.
+func (b *Bucket) forEachPageNode(fn func(*page, *node, int)) {
+	// If we have an inline page or root node then just use that.
+	if b.page != nil {
+		fn(b.page, nil, 0)
+		return
+	}
+	b._forEachPageNode(b.root, 0, fn)
+}
+
+func (b *Bucket) _forEachPageNode(pgid pgid, depth int, fn func(*page, *node, int)) {
+	var p, n = b.pageNode(pgid)
+
+	// Execute function.
+	fn(p, n, depth)
+
+	// Recursively loop over children.
+	if p != nil {
+		if (p.flags & branchPageFlag) != 0 {
+			for i := 0; i < int(p.count); i++ {
+				elem := p.branchPageElement(uint16(i))
+				b._forEachPageNode(elem.pgid, depth+1, fn)
+			}
+		}
+	} else {
+		if !n.isLeaf {
+			for _, inode := range n.inodes {
+				b._forEachPageNode(inode.pgid, depth+1, fn)
+			}
+		}
+	}
+}
+
+// spill writes all the nodes for this bucket to dirty pages.
+func (b *Bucket) spill() error {
+	// Spill all child buckets first.
+	for name, child := range b.buckets {
+		// If the child bucket is small enough and it has no child buckets then
+		// write it inline into the parent bucket's page. Otherwise spill it
+		// like a normal bucket and make the parent value a pointer to the page.
+		var value []byte
+		if child.inlineable() {
+			child.free()
+			value = child.write()
+		} else {
+			if err := child.spill(); err != nil {
+				return err
+			}
+
+			// Update the child bucket header in this bucket.
+			value = make([]byte, unsafe.Sizeof(bucket{}))
+			var bucket = (*bucket)(unsafe.Pointer(&value[0]))
+			*bucket = *child.bucket
+		}
+
+		// Skip writing the bucket if there are no materialized nodes.
+		if child.rootNode == nil {
+			continue
+		}
+
+		// Update parent node.
+		var c = b.Cursor()
+		k, _, flags := c.seek([]byte(name))
+		if !bytes.Equal([]byte(name), k) {
+			panic(fmt.Sprintf("misplaced bucket header: %x -> %x", []byte(name), k))
+		}
+		if flags&bucketLeafFlag == 0 {
+			panic(fmt.Sprintf("unexpected bucket header flag: %x", flags))
+		}
+		c.node().put([]byte(name), []byte(name), value, 0, bucketLeafFlag)
+	}
+
+	// Ignore if there's not a materialized root node.
+	if b.rootNode == nil {
+		return nil
+	}
+
+	// Spill nodes.
+	if err := b.rootNode.spill(); err != nil {
+		return err
+	}
+	b.rootNode = b.rootNode.root()
+
+	// Update the root node for this bucket.
+	if b.rootNode.pgid >= b.tx.meta.pgid {
+		panic(fmt.Sprintf("pgid (%d) above high water mark (%d)", b.rootNode.pgid, b.tx.meta.pgid))
+	}
+	b.root = b.rootNode.pgid
+
+	return nil
+}
+
+// inlineable returns true if a bucket is small enough to be written inline
+// and if it contains no subbuckets. Otherwise returns false.
+func (b *Bucket) inlineable() bool {
+	var n = b.rootNode
+
+	// Bucket must only contain a single leaf node.
+	if n == nil || !n.isLeaf {
+		return false
+	}
+
+	// Bucket is not inlineable if it contains subbuckets or if it goes beyond
+	// our threshold for inline bucket size.
+	var size = pageHeaderSize
+	for _, inode := range n.inodes {
+		size += leafPageElementSize + len(inode.key) + len(inode.value)
+
+		if inode.flags&bucketLeafFlag != 0 {
+			return false
+		} else if size > b.maxInlineBucketSize() {
+			return false
+		}
+	}
+
+	return true
+}
+
+// Returns the maximum total size of a bucket to make it a candidate for inlining.
+func (b *Bucket) maxInlineBucketSize() int {
+	return b.tx.db.pageSize / 4
+}
+
+// write allocates and writes a bucket to a byte slice.
+func (b *Bucket) write() []byte {
+	// Allocate the appropriate size.
+	var n = b.rootNode
+	var value = make([]byte, bucketHeaderSize+n.size())
+
+	// Write a bucket header.
+	var bucket = (*bucket)(unsafe.Pointer(&value[0]))
+	*bucket = *b.bucket
+
+	// Convert byte slice to a fake page and write the root node.
+	var p = (*page)(unsafe.Pointer(&value[bucketHeaderSize]))
+	n.write(p)
+
+	return value
+}
+
+// rebalance attempts to balance all nodes.
+func (b *Bucket) rebalance() {
+	for _, n := range b.nodes {
+		n.rebalance()
+	}
+	for _, child := range b.buckets {
+		child.rebalance()
+	}
+}
+
+// node creates a node from a page and associates it with a given parent.
+func (b *Bucket) node(pgid pgid, parent *node) *node {
+	_assert(b.nodes != nil, "nodes map expected")
+
+	// Retrieve node if it's already been created.
+	if n := b.nodes[pgid]; n != nil {
+		return n
+	}
+
+	// Otherwise create a node and cache it.
+	n := &node{bucket: b, parent: parent}
+	if parent == nil {
+		b.rootNode = n
+	} else {
+		parent.children = append(parent.children, n)
+	}
+
+	// Use the inline page if this is an inline bucket.
+	var p = b.page
+	if p == nil {
+		p = b.tx.page(pgid)
+	}
+
+	// Read the page into the node and cache it.
+	n.read(p)
+	b.nodes[pgid] = n
+
+	// Update statistics.
+	b.tx.stats.NodeCount++
+
+	return n
+}
+
+// free recursively frees all pages in the bucket.
+func (b *Bucket) free() {
+	if b.root == 0 {
+		return
+	}
+
+	var tx = b.tx
+	b.forEachPageNode(func(p *page, n *node, _ int) {
+		if p != nil {
+			tx.db.freelist.free(tx.meta.txid, p)
+		} else {
+			n.free()
+		}
+	})
+	b.root = 0
+}
+
+// dereference removes all references to the old mmap.
+func (b *Bucket) dereference() {
+	if b.rootNode != nil {
+		b.rootNode.root().dereference()
+	}
+
+	for _, child := range b.buckets {
+		child.dereference()
+	}
+}
+
+// pageNode returns the in-memory node, if it exists.
+// Otherwise returns the underlying page.
+func (b *Bucket) pageNode(id pgid) (*page, *node) {
+	// Inline buckets have a fake page embedded in their value so treat them
+	// differently. We'll return the rootNode (if available) or the fake page.
+	if b.root == 0 {
+		if id != 0 {
+			panic(fmt.Sprintf("inline bucket non-zero page access(2): %d != 0", id))
+		}
+		if b.rootNode != nil {
+			return nil, b.rootNode
+		}
+		return b.page, nil
+	}
+
+	// Check the node cache for non-inline buckets.
+	if b.nodes != nil {
+		if n := b.nodes[id]; n != nil {
+			return nil, n
+		}
+	}
+
+	// Finally lookup the page from the transaction if no node is materialized.
+	return b.tx.page(id), nil
+}
+
+// BucketStats records statistics about resources used by a bucket.
+type BucketStats struct {
+	// Page count statistics.
+	BranchPageN     int // number of logical branch pages
+	BranchOverflowN int // number of physical branch overflow pages
+	LeafPageN       int // number of logical leaf pages
+	LeafOverflowN   int // number of physical leaf overflow pages
+
+	// Tree statistics.
+	KeyN  int // number of keys/value pairs
+	Depth int // number of levels in B+tree
+
+	// Page size utilization.
+	BranchAlloc int // bytes allocated for physical branch pages
+	BranchInuse int // bytes actually used for branch data
+	LeafAlloc   int // bytes allocated for physical leaf pages
+	LeafInuse   int // bytes actually used for leaf data
+
+	// Bucket statistics
+	BucketN           int // total number of buckets including the top bucket
+	InlineBucketN     int // total number on inlined buckets
+	InlineBucketInuse int // bytes used for inlined buckets (also accounted for in LeafInuse)
+}
+
+func (s *BucketStats) Add(other BucketStats) {
+	s.BranchPageN += other.BranchPageN
+	s.BranchOverflowN += other.BranchOverflowN
+	s.LeafPageN += other.LeafPageN
+	s.LeafOverflowN += other.LeafOverflowN
+	s.KeyN += other.KeyN
+	if s.Depth < other.Depth {
+		s.Depth = other.Depth
+	}
+	s.BranchAlloc += other.BranchAlloc
+	s.BranchInuse += other.BranchInuse
+	s.LeafAlloc += other.LeafAlloc
+	s.LeafInuse += other.LeafInuse
+
+	s.BucketN += other.BucketN
+	s.InlineBucketN += other.InlineBucketN
+	s.InlineBucketInuse += other.InlineBucketInuse
+}
+
+// cloneBytes returns a copy of a given slice.
+func cloneBytes(v []byte) []byte {
+	var clone = make([]byte, len(v))
+	copy(clone, v)
+	return clone
+}

vendor/github.com/coreos/bbolt/cursor.go

@@ -0,0 +1,400 @@
+package bolt
+
+import (
+	"bytes"
+	"fmt"
+	"sort"
+)
+
+// Cursor represents an iterator that can traverse over all key/value pairs in a bucket in sorted order.
+// Cursors see nested buckets with value == nil.
+// Cursors can be obtained from a transaction and are valid as long as the transaction is open.
+//
+// Keys and values returned from the cursor are only valid for the life of the transaction.
+//
+// Changing data while traversing with a cursor may cause it to be invalidated
+// and return unexpected keys and/or values. You must reposition your cursor
+// after mutating data.
+type Cursor struct {
+	bucket *Bucket
+	stack  []elemRef
+}
+
+// Bucket returns the bucket that this cursor was created from.
+func (c *Cursor) Bucket() *Bucket {
+	return c.bucket
+}
+
+// First moves the cursor to the first item in the bucket and returns its key and value.
+// If the bucket is empty then a nil key and value are returned.
+// The returned key and value are only valid for the life of the transaction.
+func (c *Cursor) First() (key []byte, value []byte) {
+	_assert(c.bucket.tx.db != nil, "tx closed")
+	c.stack = c.stack[:0]
+	p, n := c.bucket.pageNode(c.bucket.root)
+	c.stack = append(c.stack, elemRef{page: p, node: n, index: 0})
+	c.first()
+
+	// If we land on an empty page then move to the next value.
+	// https://github.com/boltdb/bolt/issues/450
+	if c.stack[len(c.stack)-1].count() == 0 {
+		c.next()
+	}
+
+	k, v, flags := c.keyValue()
+	if (flags & uint32(bucketLeafFlag)) != 0 {
+		return k, nil
+	}
+	return k, v
+
+}
+
+// Last moves the cursor to the last item in the bucket and returns its key and value.
+// If the bucket is empty then a nil key and value are returned.
+// The returned key and value are only valid for the life of the transaction.
+func (c *Cursor) Last() (key []byte, value []byte) {
+	_assert(c.bucket.tx.db != nil, "tx closed")
+	c.stack = c.stack[:0]
+	p, n := c.bucket.pageNode(c.bucket.root)
+	ref := elemRef{page: p, node: n}
+	ref.index = ref.count() - 1
+	c.stack = append(c.stack, ref)
+	c.last()
+	k, v, flags := c.keyValue()
+	if (flags & uint32(bucketLeafFlag)) != 0 {
+		return k, nil
+	}
+	return k, v
+}
+
+// Next moves the cursor to the next item in the bucket and returns its key and value.
+// If the cursor is at the end of the bucket then a nil key and value are returned.
+// The returned key and value are only valid for the life of the transaction.
+func (c *Cursor) Next() (key []byte, value []byte) {
+	_assert(c.bucket.tx.db != nil, "tx closed")
+	k, v, flags := c.next()
+	if (flags & uint32(bucketLeafFlag)) != 0 {
+		return k, nil
+	}
+	return k, v
+}
+
+// Prev moves the cursor to the previous item in the bucket and returns its key and value.
+// If the cursor is at the beginning of the bucket then a nil key and value are returned.
+// The returned key and value are only valid for the life of the transaction.
+func (c *Cursor) Prev() (key []byte, value []byte) {
+	_assert(c.bucket.tx.db != nil, "tx closed")
+
+	// Attempt to move back one element until we're successful.
+	// Move up the stack as we hit the beginning of each page in our stack.
+	for i := len(c.stack) - 1; i >= 0; i-- {
+		elem := &c.stack[i]
+		if elem.index > 0 {
+			elem.index--
+			break
+		}
+		c.stack = c.stack[:i]
+	}
+
+	// If we've hit the end then return nil.
+	if len(c.stack) == 0 {
+		return nil, nil
+	}
+
+	// Move down the stack to find the last element of the last leaf under this branch.
+	c.last()
+	k, v, flags := c.keyValue()
+	if (flags & uint32(bucketLeafFlag)) != 0 {
+		return k, nil
+	}
+	return k, v
+}
+
+// Seek moves the cursor to a given key and returns it.
+// If the key does not exist then the next key is used. If no keys
+// follow, a nil key is returned.
+// The returned key and value are only valid for the life of the transaction.
+func (c *Cursor) Seek(seek []byte) (key []byte, value []byte) {
+	k, v, flags := c.seek(seek)
+
+	// If we ended up after the last element of a page then move to the next one.
+	if ref := &c.stack[len(c.stack)-1]; ref.index >= ref.count() {
+		k, v, flags = c.next()
+	}
+
+	if k == nil {
+		return nil, nil
+	} else if (flags & uint32(bucketLeafFlag)) != 0 {
+		return k, nil
+	}
+	return k, v
+}
+
+// Delete removes the current key/value under the cursor from the bucket.
+// Delete fails if current key/value is a bucket or if the transaction is not writable.
+func (c *Cursor) Delete() error {
+	if c.bucket.tx.db == nil {
+		return ErrTxClosed
+	} else if !c.bucket.Writable() {
+		return ErrTxNotWritable
+	}
+
+	key, _, flags := c.keyValue()
+	// Return an error if current value is a bucket.
+	if (flags & bucketLeafFlag) != 0 {
+		return ErrIncompatibleValue
+	}
+	c.node().del(key)
+
+	return nil
+}
+
+// seek moves the cursor to a given key and returns it.
+// If the key does not exist then the next key is used.
+func (c *Cursor) seek(seek []byte) (key []byte, value []byte, flags uint32) {
+	_assert(c.bucket.tx.db != nil, "tx closed")
+
+	// Start from root page/node and traverse to correct page.
+	c.stack = c.stack[:0]
+	c.search(seek, c.bucket.root)
+	ref := &c.stack[len(c.stack)-1]
+
+	// If the cursor is pointing to the end of page/node then return nil.
+	if ref.index >= ref.count() {
+		return nil, nil, 0
+	}
+
+	// If this is a bucket then return a nil value.
+	return c.keyValue()
+}
+
+// first moves the cursor to the first leaf element under the last page in the stack.
+func (c *Cursor) first() {
+	for {
+		// Exit when we hit a leaf page.
+		var ref = &c.stack[len(c.stack)-1]
+		if ref.isLeaf() {
+			break
+		}
+
+		// Keep adding pages pointing to the first element to the stack.
+		var pgid pgid
+		if ref.node != nil {
+			pgid = ref.node.inodes[ref.index].pgid
+		} else {
+			pgid = ref.page.branchPageElement(uint16(ref.index)).pgid
+		}
+		p, n := c.bucket.pageNode(pgid)
+		c.stack = append(c.stack, elemRef{page: p, node: n, index: 0})
+	}
+}
+
+// last moves the cursor to the last leaf element under the last page in the stack.
+func (c *Cursor) last() {
+	for {
+		// Exit when we hit a leaf page.
+		ref := &c.stack[len(c.stack)-1]
+		if ref.isLeaf() {
+			break
+		}
+
+		// Keep adding pages pointing to the last element in the stack.
+		var pgid pgid
+		if ref.node != nil {
+			pgid = ref.node.inodes[ref.index].pgid
+		} else {
+			pgid = ref.page.branchPageElement(uint16(ref.index)).pgid
+		}
+		p, n := c.bucket.pageNode(pgid)
+
+		var nextRef = elemRef{page: p, node: n}
+		nextRef.index = nextRef.count() - 1
+		c.stack = append(c.stack, nextRef)
+	}
+}
+
+// next moves to the next leaf element and returns the key and value.
+// If the cursor is at the last leaf element then it stays there and returns nil.
+func (c *Cursor) next() (key []byte, value []byte, flags uint32) {
+	for {
+		// Attempt to move over one element until we're successful.
+		// Move up the stack as we hit the end of each page in our stack.
+		var i int
+		for i = len(c.stack) - 1; i >= 0; i-- {
+			elem := &c.stack[i]
+			if elem.index < elem.count()-1 {
+				elem.index++
+				break
+			}
+		}
+
+		// If we've hit the root page then stop and return. This will leave the
+		// cursor on the last element of the last page.
+		if i == -1 {
+			return nil, nil, 0
+		}
+
+		// Otherwise start from where we left off in the stack and find the
+		// first element of the first leaf page.
+		c.stack = c.stack[:i+1]
+		c.first()
+
+		// If this is an empty page then restart and move back up the stack.
+		// https://github.com/boltdb/bolt/issues/450
+		if c.stack[len(c.stack)-1].count() == 0 {
+			continue
+		}
+
+		return c.keyValue()
+	}
+}
+
+// search recursively performs a binary search against a given page/node until it finds a given key.
+func (c *Cursor) search(key []byte, pgid pgid) {
+	p, n := c.bucket.pageNode(pgid)
+	if p != nil && (p.flags&(branchPageFlag|leafPageFlag)) == 0 {
+		panic(fmt.Sprintf("invalid page type: %d: %x", p.id, p.flags))
+	}
+	e := elemRef{page: p, node: n}
+	c.stack = append(c.stack, e)
+
+	// If we're on a leaf page/node then find the specific node.
+	if e.isLeaf() {
+		c.nsearch(key)
+		return
+	}
+
+	if n != nil {
+		c.searchNode(key, n)
+		return
+	}
+	c.searchPage(key, p)
+}
+
+func (c *Cursor) searchNode(key []byte, n *node) {
+	var exact bool
+	index := sort.Search(len(n.inodes), func(i int) bool {
+		// TODO(benbjohnson): Optimize this range search. It's a bit hacky right now.
+		// sort.Search() finds the lowest index where f() != -1 but we need the highest index.
+		ret := bytes.Compare(n.inodes[i].key, key)
+		if ret == 0 {
+			exact = true
+		}
+		return ret != -1
+	})
+	if !exact && index > 0 {
+		index--
+	}
+	c.stack[len(c.stack)-1].index = index
+
+	// Recursively search to the next page.
+	c.search(key, n.inodes[index].pgid)
+}
+
+func (c *Cursor) searchPage(key []byte, p *page) {
+	// Binary search for the correct range.
+	inodes := p.branchPageElements()
+
+	var exact bool
+	index := sort.Search(int(p.count), func(i int) bool {
+		// TODO(benbjohnson): Optimize this range search. It's a bit hacky right now.
+		// sort.Search() finds the lowest index where f() != -1 but we need the highest index.
+		ret := bytes.Compare(inodes[i].key(), key)
+		if ret == 0 {
+			exact = true
+		}
+		return ret != -1
+	})
+	if !exact && index > 0 {
+		index--
+	}
+	c.stack[len(c.stack)-1].index = index
+
+	// Recursively search to the next page.
+	c.search(key, inodes[index].pgid)
+}
+
+// nsearch searches the leaf node on the top of the stack for a key.
+func (c *Cursor) nsearch(key []byte) {
+	e := &c.stack[len(c.stack)-1]
+	p, n := e.page, e.node
+
+	// If we have a node then search its inodes.
+	if n != nil {
+		index := sort.Search(len(n.inodes), func(i int) bool {
+			return bytes.Compare(n.inodes[i].key, key) != -1
+		})
+		e.index = index
+		return
+	}
+
+	// If we have a page then search its leaf elements.
+	inodes := p.leafPageElements()
+	index := sort.Search(int(p.count), func(i int) bool {
+		return bytes.Compare(inodes[i].key(), key) != -1
+	})
+	e.index = index
+}
+
+// keyValue returns the key and value of the current leaf element.
+func (c *Cursor) keyValue() ([]byte, []byte, uint32) {
+	ref := &c.stack[len(c.stack)-1]
+	if ref.count() == 0 || ref.index >= ref.count() {
+		return nil, nil, 0
+	}
+
+	// Retrieve value from node.
+	if ref.node != nil {
+		inode := &ref.node.inodes[ref.index]
+		return inode.key, inode.value, inode.flags
+	}
+
+	// Or retrieve value from page.
+	elem := ref.page.leafPageElement(uint16(ref.index))
+	return elem.key(), elem.value(), elem.flags
+}
+
+// node returns the node that the cursor is currently positioned on.
+func (c *Cursor) node() *node {
+	_assert(len(c.stack) > 0, "accessing a node with a zero-length cursor stack")
+
+	// If the top of the stack is a leaf node then just return it.
+	if ref := &c.stack[len(c.stack)-1]; ref.node != nil && ref.isLeaf() {
+		return ref.node
+	}
+
+	// Start from root and traverse down the hierarchy.
+	var n = c.stack[0].node
+	if n == nil {
+		n = c.bucket.node(c.stack[0].page.id, nil)
+	}
+	for _, ref := range c.stack[:len(c.stack)-1] {
+		_assert(!n.isLeaf, "expected branch node")
+		n = n.childAt(int(ref.index))
+	}
+	_assert(n.isLeaf, "expected leaf node")
+	return n
+}
+
+// elemRef represents a reference to an element on a given page/node.
+type elemRef struct {
+	page  *page
+	node  *node
+	index int
+}
+
+// isLeaf returns whether the ref is pointing at a leaf page/node.
+func (r *elemRef) isLeaf() bool {
+	if r.node != nil {
+		return r.node.isLeaf
+	}
+	return (r.page.flags & leafPageFlag) != 0
+}
+
+// count returns the number of inodes or page elements.
+func (r *elemRef) count() int {
+	if r.node != nil {
+		return len(r.node.inodes)
+	}
+	return int(r.page.count)
+}

vendor/github.com/coreos/bbolt/doc.go

@@ -0,0 +1,44 @@
+/*
+Package bolt implements a low-level key/value store in pure Go. It supports
+fully serializable transactions, ACID semantics, and lock-free MVCC with
+multiple readers and a single writer. Bolt can be used for projects that
+want a simple data store without the need to add large dependencies such as
+Postgres or MySQL.
+
+Bolt is a single-level, zero-copy, B+tree data store. This means that Bolt is
+optimized for fast read access and does not require recovery in the event of a
+system crash. Transactions which have not finished committing will simply be
+rolled back in the event of a crash.
+
+The design of Bolt is based on Howard Chu's LMDB database project.
+
+Bolt currently works on Windows, Mac OS X, and Linux.
+
+
+Basics
+
+There are only a few types in Bolt: DB, Bucket, Tx, and Cursor. The DB is
+a collection of buckets and is represented by a single file on disk. A bucket is
+a collection of unique keys that are associated with values.
+
+Transactions provide either read-only or read-write access to the database.
+Read-only transactions can retrieve key/value pairs and can use Cursors to
+iterate over the dataset sequentially. Read-write transactions can create and
+delete buckets and can insert and remove keys. Only one read-write transaction
+is allowed at a time.
+
+
+Caveats
+
+The database uses a read-only, memory-mapped data file to ensure that
+applications cannot corrupt the database, however, this means that keys and
+values returned from Bolt cannot be changed. Writing to a read-only byte slice
+will cause Go to panic.
+
+Keys and values retrieved from the database are only valid for the life of
+the transaction. When used outside the transaction, these byte slices can
+point to different data or can point to invalid memory which will cause a panic.
+
+
+*/
+package bolt

vendor/github.com/coreos/bbolt/errors.go

@@ -0,0 +1,71 @@
+package bolt
+
+import "errors"
+
+// These errors can be returned when opening or calling methods on a DB.
+var (
+	// ErrDatabaseNotOpen is returned when a DB instance is accessed before it
+	// is opened or after it is closed.
+	ErrDatabaseNotOpen = errors.New("database not open")
+
+	// ErrDatabaseOpen is returned when opening a database that is
+	// already open.
+	ErrDatabaseOpen = errors.New("database already open")
+
+	// ErrInvalid is returned when both meta pages on a database are invalid.
+	// This typically occurs when a file is not a bolt database.
+	ErrInvalid = errors.New("invalid database")
+
+	// ErrVersionMismatch is returned when the data file was created with a
+	// different version of Bolt.
+	ErrVersionMismatch = errors.New("version mismatch")
+
+	// ErrChecksum is returned when either meta page checksum does not match.
+	ErrChecksum = errors.New("checksum error")
+
+	// ErrTimeout is returned when a database cannot obtain an exclusive lock
+	// on the data file after the timeout passed to Open().
+	ErrTimeout = errors.New("timeout")
+)
+
+// These errors can occur when beginning or committing a Tx.
+var (
+	// ErrTxNotWritable is returned when performing a write operation on a
+	// read-only transaction.
+	ErrTxNotWritable = errors.New("tx not writable")
+
+	// ErrTxClosed is returned when committing or rolling back a transaction
+	// that has already been committed or rolled back.
+	ErrTxClosed = errors.New("tx closed")
+
+	// ErrDatabaseReadOnly is returned when a mutating transaction is started on a
+	// read-only database.
+	ErrDatabaseReadOnly = errors.New("database is in read-only mode")
+)
+
+// These errors can occur when putting or deleting a value or a bucket.
+var (
+	// ErrBucketNotFound is returned when trying to access a bucket that has
+	// not been created yet.
+	ErrBucketNotFound = errors.New("bucket not found")
+
+	// ErrBucketExists is returned when creating a bucket that already exists.
+	ErrBucketExists = errors.New("bucket already exists")
+
+	// ErrBucketNameRequired is returned when creating a bucket with a blank name.
+	ErrBucketNameRequired = errors.New("bucket name required")
+
+	// ErrKeyRequired is returned when inserting a zero-length key.
+	ErrKeyRequired = errors.New("key required")
+
+	// ErrKeyTooLarge is returned when inserting a key that is larger than MaxKeySize.
+	ErrKeyTooLarge = errors.New("key too large")
+
+	// ErrValueTooLarge is returned when inserting a value that is larger than MaxValueSize.
+	ErrValueTooLarge = errors.New("value too large")
+
+	// ErrIncompatibleValue is returned when trying create or delete a bucket
+	// on an existing non-bucket key or when trying to create or delete a
+	// non-bucket key on an existing bucket key.
+	ErrIncompatibleValue = errors.New("incompatible value")
+)

vendor/github.com/coreos/bbolt/freelist.go

@@ -0,0 +1,330 @@
+package bolt
+
+import (
+	"fmt"
+	"sort"
+	"unsafe"
+)
+
+// txPending holds a list of pgids and corresponding allocation txns
+// that are pending to be freed.
+type txPending struct {
+	ids              []pgid
+	alloctx          []txid // txids allocating the ids
+	lastReleaseBegin txid   // beginning txid of last matching releaseRange
+}
+
+// freelist represents a list of all pages that are available for allocation.
+// It also tracks pages that have been freed but are still in use by open transactions.
+type freelist struct {
+	ids     []pgid              // all free and available free page ids.
+	allocs  map[pgid]txid       // mapping of txid that allocated a pgid.
+	pending map[txid]*txPending // mapping of soon-to-be free page ids by tx.
+	cache   map[pgid]bool       // fast lookup of all free and pending page ids.
+}
+
+// newFreelist returns an empty, initialized freelist.
+func newFreelist() *freelist {
+	return &freelist{
+		allocs:  make(map[pgid]txid),
+		pending: make(map[txid]*txPending),
+		cache:   make(map[pgid]bool),
+	}
+}
+
+// size returns the size of the page after serialization.
+func (f *freelist) size() int {
+	n := f.count()
+	if n >= 0xFFFF {
+		// The first element will be used to store the count. See freelist.write.
+		n++
+	}
+	return pageHeaderSize + (int(unsafe.Sizeof(pgid(0))) * n)
+}
+
+// count returns count of pages on the freelist
+func (f *freelist) count() int {
+	return f.free_count() + f.pending_count()
+}
+
+// free_count returns count of free pages
+func (f *freelist) free_count() int {
+	return len(f.ids)
+}
+
+// pending_count returns count of pending pages
+func (f *freelist) pending_count() int {
+	var count int
+	for _, txp := range f.pending {
+		count += len(txp.ids)
+	}
+	return count
+}
+
+// copyall copies into dst a list of all free ids and all pending ids in one sorted list.
+// f.count returns the minimum length required for dst.
+func (f *freelist) copyall(dst []pgid) {
+	m := make(pgids, 0, f.pending_count())
+	for _, txp := range f.pending {
+		m = append(m, txp.ids...)
+	}
+	sort.Sort(m)
+	mergepgids(dst, f.ids, m)
+}
+
+// allocate returns the starting page id of a contiguous list of pages of a given size.
+// If a contiguous block cannot be found then 0 is returned.
+func (f *freelist) allocate(txid txid, n int) pgid {
+	if len(f.ids) == 0 {
+		return 0
+	}
+
+	var initial, previd pgid
+	for i, id := range f.ids {
+		if id <= 1 {
+			panic(fmt.Sprintf("invalid page allocation: %d", id))
+		}
+
+		// Reset initial page if this is not contiguous.
+		if previd == 0 || id-previd != 1 {
+			initial = id
+		}
+
+		// If we found a contiguous block then remove it and return it.
+		if (id-initial)+1 == pgid(n) {
+			// If we're allocating off the beginning then take the fast path
+			// and just adjust the existing slice. This will use extra memory
+			// temporarily but the append() in free() will realloc the slice
+			// as is necessary.
+			if (i + 1) == n {
+				f.ids = f.ids[i+1:]
+			} else {
+				copy(f.ids[i-n+1:], f.ids[i+1:])
+				f.ids = f.ids[:len(f.ids)-n]
+			}
+
+			// Remove from the free cache.
+			for i := pgid(0); i < pgid(n); i++ {
+				delete(f.cache, initial+i)
+			}
+			f.allocs[initial] = txid
+			return initial
+		}
+
+		previd = id
+	}
+	return 0
+}
+
+// free releases a page and its overflow for a given transaction id.
+// If the page is already free then a panic will occur.
+func (f *freelist) free(txid txid, p *page) {
+	if p.id <= 1 {
+		panic(fmt.Sprintf("cannot free page 0 or 1: %d", p.id))
+	}
+
+	// Free page and all its overflow pages.
+	txp := f.pending[txid]
+	if txp == nil {
+		txp = &txPending{}
+		f.pending[txid] = txp
+	}
+	allocTxid, ok := f.allocs[p.id]
+	if ok {
+		delete(f.allocs, p.id)
+	} else if (p.flags & (freelistPageFlag | metaPageFlag)) != 0 {
+		// Safe to claim txid as allocating since these types are private to txid.
+		allocTxid = txid
+	}
+
+	for id := p.id; id <= p.id+pgid(p.overflow); id++ {
+		// Verify that page is not already free.
+		if f.cache[id] {
+			panic(fmt.Sprintf("page %d already freed", id))
+		}
+		// Add to the freelist and cache.
+		txp.ids = append(txp.ids, id)
+		txp.alloctx = append(txp.alloctx, allocTxid)
+		f.cache[id] = true
+	}
+}
+
+// release moves all page ids for a transaction id (or older) to the freelist.
+func (f *freelist) release(txid txid) {
+	m := make(pgids, 0)
+	for tid, txp := range f.pending {
+		if tid <= txid {
+			// Move transaction's pending pages to the available freelist.
+			// Don't remove from the cache since the page is still free.
+			m = append(m, txp.ids...)
+			delete(f.pending, tid)
+		}
+	}
+	sort.Sort(m)
+	f.ids = pgids(f.ids).merge(m)
+}
+
+// releaseRange moves pending pages allocated within an extent [begin,end] to the free list.
+func (f *freelist) releaseRange(begin, end txid) {
+	if begin > end {
+		return
+	}
+	var m pgids
+	for tid, txp := range f.pending {
+		if tid < begin || tid > end {
+			continue
+		}
+		// Don't recompute freed pages if ranges haven't updated.
+		if txp.lastReleaseBegin == begin {
+			continue
+		}
+		for i := 0; i < len(txp.ids); i++ {
+			if atx := txp.alloctx[i]; atx < begin || atx > end {
+				continue
+			}
+			m = append(m, txp.ids[i])
+			txp.ids[i] = txp.ids[len(txp.ids)-1]
+			txp.ids = txp.ids[:len(txp.ids)-1]
+			txp.alloctx[i] = txp.alloctx[len(txp.alloctx)-1]
+			txp.alloctx = txp.alloctx[:len(txp.alloctx)-1]
+			i--
+		}
+		txp.lastReleaseBegin = begin
+		if len(txp.ids) == 0 {
+			delete(f.pending, tid)
+		}
+	}
+	sort.Sort(m)
+	f.ids = pgids(f.ids).merge(m)
+}
+
+// rollback removes the pages from a given pending tx.
+func (f *freelist) rollback(txid txid) {
+	// Remove page ids from cache.
+	txp := f.pending[txid]
+	if txp == nil {
+		return
+	}
+	var m pgids
+	for i, pgid := range txp.ids {
+		delete(f.cache, pgid)
+		tx := txp.alloctx[i]
+		if tx == 0 {
+			continue
+		}
+		if tx != txid {
+			// Pending free aborted; restore page back to alloc list.
+			f.allocs[pgid] = tx
+		} else {
+			// Freed page was allocated by this txn; OK to throw away.
+			m = append(m, pgid)
+		}
+	}
+	// Remove pages from pending list and mark as free if allocated by txid.
+	delete(f.pending, txid)
+	sort.Sort(m)
+	f.ids = pgids(f.ids).merge(m)
+}
+
+// freed returns whether a given page is in the free list.
+func (f *freelist) freed(pgid pgid) bool {
+	return f.cache[pgid]
+}
+
+// read initializes the freelist from a freelist page.
+func (f *freelist) read(p *page) {
+	// If the page.count is at the max uint16 value (64k) then it's considered
+	// an overflow and the size of the freelist is stored as the first element.
+	idx, count := 0, int(p.count)
+	if count == 0xFFFF {
+		idx = 1
+		count = int(((*[maxAllocSize]pgid)(unsafe.Pointer(&p.ptr)))[0])
+	}
+
+	// Copy the list of page ids from the freelist.
+	if count == 0 {
+		f.ids = nil
+	} else {
+		ids := ((*[maxAllocSize]pgid)(unsafe.Pointer(&p.ptr)))[idx : idx+count]
+		f.ids = make([]pgid, len(ids))
+		copy(f.ids, ids)
+
+		// Make sure they're sorted.
+		sort.Sort(pgids(f.ids))
+	}
+
+	// Rebuild the page cache.
+	f.reindex()
+}
+
+// read initializes the freelist from a given list of ids.
+func (f *freelist) readIDs(ids []pgid) {
+	f.ids = ids
+	f.reindex()
+}
+
+// write writes the page ids onto a freelist page. All free and pending ids are
+// saved to disk since in the event of a program crash, all pending ids will
+// become free.
+func (f *freelist) write(p *page) error {
+	// Combine the old free pgids and pgids waiting on an open transaction.
+
+	// Update the header flag.
+	p.flags |= freelistPageFlag
+
+	// The page.count can only hold up to 64k elements so if we overflow that
+	// number then we handle it by putting the size in the first element.
+	lenids := f.count()
+	if lenids == 0 {
+		p.count = uint16(lenids)
+	} else if lenids < 0xFFFF {
+		p.count = uint16(lenids)
+		f.copyall(((*[maxAllocSize]pgid)(unsafe.Pointer(&p.ptr)))[:])
+	} else {
+		p.count = 0xFFFF
+		((*[maxAllocSize]pgid)(unsafe.Pointer(&p.ptr)))[0] = pgid(lenids)
+		f.copyall(((*[maxAllocSize]pgid)(unsafe.Pointer(&p.ptr)))[1:])
+	}
+
+	return nil
+}
+
+// reload reads the freelist from a page and filters out pending items.
+func (f *freelist) reload(p *page) {
+	f.read(p)
+
+	// Build a cache of only pending pages.
+	pcache := make(map[pgid]bool)
+	for _, txp := range f.pending {
+		for _, pendingID := range txp.ids {
+			pcache[pendingID] = true
+		}
+	}
+
+	// Check each page in the freelist and build a new available freelist
+	// with any pages not in the pending lists.
+	var a []pgid
+	for _, id := range f.ids {
+		if !pcache[id] {
+			a = append(a, id)
+		}
+	}
+	f.ids = a
+
+	// Once the available list is rebuilt then rebuild the free cache so that
+	// it includes the available and pending free pages.
+	f.reindex()
+}
+
+// reindex rebuilds the free cache based on available and pending free lists.
+func (f *freelist) reindex() {
+	f.cache = make(map[pgid]bool, len(f.ids))
+	for _, id := range f.ids {
+		f.cache[id] = true
+	}
+	for _, txp := range f.pending {
+		for _, pendingID := range txp.ids {
+			f.cache[pendingID] = true
+		}
+	}
+}

vendor/github.com/coreos/bbolt/node.go

@@ -0,0 +1,604 @@
+package bolt
+
+import (
+	"bytes"
+	"fmt"
+	"sort"
+	"unsafe"
+)
+
+// node represents an in-memory, deserialized page.
+type node struct {
+	bucket     *Bucket
+	isLeaf     bool
+	unbalanced bool
+	spilled    bool
+	key        []byte
+	pgid       pgid
+	parent     *node
+	children   nodes
+	inodes     inodes
+}
+
+// root returns the top-level node this node is attached to.
+func (n *node) root() *node {
+	if n.parent == nil {
+		return n
+	}
+	return n.parent.root()
+}
+
+// minKeys returns the minimum number of inodes this node should have.
+func (n *node) minKeys() int {
+	if n.isLeaf {
+		return 1
+	}
+	return 2
+}
+
+// size returns the size of the node after serialization.
+func (n *node) size() int {
+	sz, elsz := pageHeaderSize, n.pageElementSize()
+	for i := 0; i < len(n.inodes); i++ {
+		item := &n.inodes[i]
+		sz += elsz + len(item.key) + len(item.value)
+	}
+	return sz
+}
+
+// sizeLessThan returns true if the node is less than a given size.
+// This is an optimization to avoid calculating a large node when we only need
+// to know if it fits inside a certain page size.
+func (n *node) sizeLessThan(v int) bool {
+	sz, elsz := pageHeaderSize, n.pageElementSize()
+	for i := 0; i < len(n.inodes); i++ {
+		item := &n.inodes[i]
+		sz += elsz + len(item.key) + len(item.value)
+		if sz >= v {
+			return false
+		}
+	}
+	return true
+}
+
+// pageElementSize returns the size of each page element based on the type of node.
+func (n *node) pageElementSize() int {
+	if n.isLeaf {
+		return leafPageElementSize
+	}
+	return branchPageElementSize
+}
+
+// childAt returns the child node at a given index.
+func (n *node) childAt(index int) *node {
+	if n.isLeaf {
+		panic(fmt.Sprintf("invalid childAt(%d) on a leaf node", index))
+	}
+	return n.bucket.node(n.inodes[index].pgid, n)
+}
+
+// childIndex returns the index of a given child node.
+func (n *node) childIndex(child *node) int {
+	index := sort.Search(len(n.inodes), func(i int) bool { return bytes.Compare(n.inodes[i].key, child.key) != -1 })
+	return index
+}
+
+// numChildren returns the number of children.
+func (n *node) numChildren() int {
+	return len(n.inodes)
+}
+
+// nextSibling returns the next node with the same parent.
+func (n *node) nextSibling() *node {
+	if n.parent == nil {
+		return nil
+	}
+	index := n.parent.childIndex(n)
+	if index >= n.parent.numChildren()-1 {
+		return nil
+	}
+	return n.parent.childAt(index + 1)
+}
+
+// prevSibling returns the previous node with the same parent.
+func (n *node) prevSibling() *node {
+	if n.parent == nil {
+		return nil
+	}
+	index := n.parent.childIndex(n)
+	if index == 0 {
+		return nil
+	}
+	return n.parent.childAt(index - 1)
+}
+
+// put inserts a key/value.
+func (n *node) put(oldKey, newKey, value []byte, pgid pgid, flags uint32) {
+	if pgid >= n.bucket.tx.meta.pgid {
+		panic(fmt.Sprintf("pgid (%d) above high water mark (%d)", pgid, n.bucket.tx.meta.pgid))
+	} else if len(oldKey) <= 0 {
+		panic("put: zero-length old key")
+	} else if len(newKey) <= 0 {
+		panic("put: zero-length new key")
+	}
+
+	// Find insertion index.
+	index := sort.Search(len(n.inodes), func(i int) bool { return bytes.Compare(n.inodes[i].key, oldKey) != -1 })
+
+	// Add capacity and shift nodes if we don't have an exact match and need to insert.
+	exact := (len(n.inodes) > 0 && index < len(n.inodes) && bytes.Equal(n.inodes[index].key, oldKey))
+	if !exact {
+		n.inodes = append(n.inodes, inode{})
+		copy(n.inodes[index+1:], n.inodes[index:])
+	}
+
+	inode := &n.inodes[index]
+	inode.flags = flags
+	inode.key = newKey
+	inode.value = value
+	inode.pgid = pgid
+	_assert(len(inode.key) > 0, "put: zero-length inode key")
+}
+
+// del removes a key from the node.
+func (n *node) del(key []byte) {
+	// Find index of key.
+	index := sort.Search(len(n.inodes), func(i int) bool { return bytes.Compare(n.inodes[i].key, key) != -1 })
+
+	// Exit if the key isn't found.
+	if index >= len(n.inodes) || !bytes.Equal(n.inodes[index].key, key) {
+		return
+	}
+
+	// Delete inode from the node.
+	n.inodes = append(n.inodes[:index], n.inodes[index+1:]...)
+
+	// Mark the node as needing rebalancing.
+	n.unbalanced = true
+}
+
+// read initializes the node from a page.
+func (n *node) read(p *page) {
+	n.pgid = p.id
+	n.isLeaf = ((p.flags & leafPageFlag) != 0)
+	n.inodes = make(inodes, int(p.count))
+
+	for i := 0; i < int(p.count); i++ {
+		inode := &n.inodes[i]
+		if n.isLeaf {
+			elem := p.leafPageElement(uint16(i))
+			inode.flags = elem.flags
+			inode.key = elem.key()
+			inode.value = elem.value()
+		} else {
+			elem := p.branchPageElement(uint16(i))
+			inode.pgid = elem.pgid
+			inode.key = elem.key()
+		}
+		_assert(len(inode.key) > 0, "read: zero-length inode key")
+	}
+
+	// Save first key so we can find the node in the parent when we spill.
+	if len(n.inodes) > 0 {
+		n.key = n.inodes[0].key
+		_assert(len(n.key) > 0, "read: zero-length node key")
+	} else {
+		n.key = nil
+	}
+}
+
+// write writes the items onto one or more pages.
+func (n *node) write(p *page) {
+	// Initialize page.
+	if n.isLeaf {
+		p.flags |= leafPageFlag
+	} else {
+		p.flags |= branchPageFlag
+	}
+
+	if len(n.inodes) >= 0xFFFF {
+		panic(fmt.Sprintf("inode overflow: %d (pgid=%d)", len(n.inodes), p.id))
+	}
+	p.count = uint16(len(n.inodes))
+
+	// Stop here if there are no items to write.
+	if p.count == 0 {
+		return
+	}
+
+	// Loop over each item and write it to the page.
+	b := (*[maxAllocSize]byte)(unsafe.Pointer(&p.ptr))[n.pageElementSize()*len(n.inodes):]
+	for i, item := range n.inodes {
+		_assert(len(item.key) > 0, "write: zero-length inode key")
+
+		// Write the page element.
+		if n.isLeaf {
+			elem := p.leafPageElement(uint16(i))
+			elem.pos = uint32(uintptr(unsafe.Pointer(&b[0])) - uintptr(unsafe.Pointer(elem)))
+			elem.flags = item.flags
+			elem.ksize = uint32(len(item.key))
+			elem.vsize = uint32(len(item.value))
+		} else {
+			elem := p.branchPageElement(uint16(i))
+			elem.pos = uint32(uintptr(unsafe.Pointer(&b[0])) - uintptr(unsafe.Pointer(elem)))
+			elem.ksize = uint32(len(item.key))
+			elem.pgid = item.pgid
+			_assert(elem.pgid != p.id, "write: circular dependency occurred")
+		}
+
+		// If the length of key+value is larger than the max allocation size
+		// then we need to reallocate the byte array pointer.
+		//
+		// See: https://github.com/boltdb/bolt/pull/335
+		klen, vlen := len(item.key), len(item.value)
+		if len(b) < klen+vlen {
+			b = (*[maxAllocSize]byte)(unsafe.Pointer(&b[0]))[:]
+		}
+
+		// Write data for the element to the end of the page.
+		copy(b[0:], item.key)
+		b = b[klen:]
+		copy(b[0:], item.value)
+		b = b[vlen:]
+	}
+
+	// DEBUG ONLY: n.dump()
+}
+
+// split breaks up a node into multiple smaller nodes, if appropriate.
+// This should only be called from the spill() function.
+func (n *node) split(pageSize int) []*node {
+	var nodes []*node
+
+	node := n
+	for {
+		// Split node into two.
+		a, b := node.splitTwo(pageSize)
+		nodes = append(nodes, a)
+
+		// If we can't split then exit the loop.
+		if b == nil {
+			break
+		}
+
+		// Set node to b so it gets split on the next iteration.
+		node = b
+	}
+
+	return nodes
+}
+
+// splitTwo breaks up a node into two smaller nodes, if appropriate.
+// This should only be called from the split() function.
+func (n *node) splitTwo(pageSize int) (*node, *node) {
+	// Ignore the split if the page doesn't have at least enough nodes for
+	// two pages or if the nodes can fit in a single page.
+	if len(n.inodes) <= (minKeysPerPage*2) || n.sizeLessThan(pageSize) {
+		return n, nil
+	}
+
+	// Determine the threshold before starting a new node.
+	var fillPercent = n.bucket.FillPercent
+	if fillPercent < minFillPercent {
+		fillPercent = minFillPercent
+	} else if fillPercent > maxFillPercent {
+		fillPercent = maxFillPercent
+	}
+	threshold := int(float64(pageSize) * fillPercent)
+
+	// Determine split position and sizes of the two pages.
+	splitIndex, _ := n.splitIndex(threshold)
+
+	// Split node into two separate nodes.
+	// If there's no parent then we'll need to create one.
+	if n.parent == nil {
+		n.parent = &node{bucket: n.bucket, children: []*node{n}}
+	}
+
+	// Create a new node and add it to the parent.
+	next := &node{bucket: n.bucket, isLeaf: n.isLeaf, parent: n.parent}
+	n.parent.children = append(n.parent.children, next)
+
+	// Split inodes across two nodes.
+	next.inodes = n.inodes[splitIndex:]
+	n.inodes = n.inodes[:splitIndex]
+
+	// Update the statistics.
+	n.bucket.tx.stats.Split++
+
+	return n, next
+}
+
+// splitIndex finds the position where a page will fill a given threshold.
+// It returns the index as well as the size of the first page.
+// This is only be called from split().
+func (n *node) splitIndex(threshold int) (index, sz int) {
+	sz = pageHeaderSize
+
+	// Loop until we only have the minimum number of keys required for the second page.
+	for i := 0; i < len(n.inodes)-minKeysPerPage; i++ {
+		index = i
+		inode := n.inodes[i]
+		elsize := n.pageElementSize() + len(inode.key) + len(inode.value)
+
+		// If we have at least the minimum number of keys and adding another
+		// node would put us over the threshold then exit and return.
+		if i >= minKeysPerPage && sz+elsize > threshold {
+			break
+		}
+
+		// Add the element size to the total size.
+		sz += elsize
+	}
+
+	return
+}
+
+// spill writes the nodes to dirty pages and splits nodes as it goes.
+// Returns an error if dirty pages cannot be allocated.
+func (n *node) spill() error {
+	var tx = n.bucket.tx
+	if n.spilled {
+		return nil
+	}
+
+	// Spill child nodes first. Child nodes can materialize sibling nodes in
+	// the case of split-merge so we cannot use a range loop. We have to check
+	// the children size on every loop iteration.
+	sort.Sort(n.children)
+	for i := 0; i < len(n.children); i++ {
+		if err := n.children[i].spill(); err != nil {
+			return err
+		}
+	}
+
+	// We no longer need the child list because it's only used for spill tracking.
+	n.children = nil
+
+	// Split nodes into appropriate sizes. The first node will always be n.
+	var nodes = n.split(tx.db.pageSize)
+	for _, node := range nodes {
+		// Add node's page to the freelist if it's not new.
+		if node.pgid > 0 {
+			tx.db.freelist.free(tx.meta.txid, tx.page(node.pgid))
+			node.pgid = 0
+		}
+
+		// Allocate contiguous space for the node.
+		p, err := tx.allocate((node.size() + tx.db.pageSize - 1) / tx.db.pageSize)
+		if err != nil {
+			return err
+		}
+
+		// Write the node.
+		if p.id >= tx.meta.pgid {
+			panic(fmt.Sprintf("pgid (%d) above high water mark (%d)", p.id, tx.meta.pgid))
+		}
+		node.pgid = p.id
+		node.write(p)
+		node.spilled = true
+
+		// Insert into parent inodes.
+		if node.parent != nil {
+			var key = node.key
+			if key == nil {
+				key = node.inodes[0].key
+			}
+
+			node.parent.put(key, node.inodes[0].key, nil, node.pgid, 0)
+			node.key = node.inodes[0].key
+			_assert(len(node.key) > 0, "spill: zero-length node key")
+		}
+
+		// Update the statistics.
+		tx.stats.Spill++
+	}
+
+	// If the root node split and created a new root then we need to spill that
+	// as well. We'll clear out the children to make sure it doesn't try to respill.
+	if n.parent != nil && n.parent.pgid == 0 {
+		n.children = nil
+		return n.parent.spill()
+	}
+
+	return nil
+}
+
+// rebalance attempts to combine the node with sibling nodes if the node fill
+// size is below a threshold or if there are not enough keys.
+func (n *node) rebalance() {
+	if !n.unbalanced {
+		return
+	}
+	n.unbalanced = false
+
+	// Update statistics.
+	n.bucket.tx.stats.Rebalance++
+
+	// Ignore if node is above threshold (25%) and has enough keys.
+	var threshold = n.bucket.tx.db.pageSize / 4
+	if n.size() > threshold && len(n.inodes) > n.minKeys() {
+		return
+	}
+
+	// Root node has special handling.
+	if n.parent == nil {
+		// If root node is a branch and only has one node then collapse it.
+		if !n.isLeaf && len(n.inodes) == 1 {
+			// Move root's child up.
+			child := n.bucket.node(n.inodes[0].pgid, n)
+			n.isLeaf = child.isLeaf
+			n.inodes = child.inodes[:]
+			n.children = child.children
+
+			// Reparent all child nodes being moved.
+			for _, inode := range n.inodes {
+				if child, ok := n.bucket.nodes[inode.pgid]; ok {
+					child.parent = n
+				}
+			}
+
+			// Remove old child.
+			child.parent = nil
+			delete(n.bucket.nodes, child.pgid)
+			child.free()
+		}
+
+		return
+	}
+
+	// If node has no keys then just remove it.
+	if n.numChildren() == 0 {
+		n.parent.del(n.key)
+		n.parent.removeChild(n)
+		delete(n.bucket.nodes, n.pgid)
+		n.free()
+		n.parent.rebalance()
+		return
+	}
+
+	_assert(n.parent.numChildren() > 1, "parent must have at least 2 children")
+
+	// Destination node is right sibling if idx == 0, otherwise left sibling.
+	var target *node
+	var useNextSibling = (n.parent.childIndex(n) == 0)
+	if useNextSibling {
+		target = n.nextSibling()
+	} else {
+		target = n.prevSibling()
+	}
+
+	// If both this node and the target node are too small then merge them.
+	if useNextSibling {
+		// Reparent all child nodes being moved.
+		for _, inode := range target.inodes {
+			if child, ok := n.bucket.nodes[inode.pgid]; ok {
+				child.parent.removeChild(child)
+				child.parent = n
+				child.parent.children = append(child.parent.children, child)
+			}
+		}
+
+		// Copy over inodes from target and remove target.
+		n.inodes = append(n.inodes, target.inodes...)
+		n.parent.del(target.key)
+		n.parent.removeChild(target)
+		delete(n.bucket.nodes, target.pgid)
+		target.free()
+	} else {
+		// Reparent all child nodes being moved.
+		for _, inode := range n.inodes {
+			if child, ok := n.bucket.nodes[inode.pgid]; ok {
+				child.parent.removeChild(child)
+				child.parent = target
+				child.parent.children = append(child.parent.children, child)
+			}
+		}
+
+		// Copy over inodes to target and remove node.
+		target.inodes = append(target.inodes, n.inodes...)
+		n.parent.del(n.key)
+		n.parent.removeChild(n)
+		delete(n.bucket.nodes, n.pgid)
+		n.free()
+	}
+
+	// Either this node or the target node was deleted from the parent so rebalance it.
+	n.parent.rebalance()
+}
+
+// removes a node from the list of in-memory children.
+// This does not affect the inodes.
+func (n *node) removeChild(target *node) {
+	for i, child := range n.children {
+		if child == target {
+			n.children = append(n.children[:i], n.children[i+1:]...)
+			return
+		}
+	}
+}
+
+// dereference causes the node to copy all its inode key/value references to heap memory.
+// This is required when the mmap is reallocated so inodes are not pointing to stale data.
+func (n *node) dereference() {
+	if n.key != nil {
+		key := make([]byte, len(n.key))
+		copy(key, n.key)
+		n.key = key
+		_assert(n.pgid == 0 || len(n.key) > 0, "dereference: zero-length node key on existing node")
+	}
+
+	for i := range n.inodes {
+		inode := &n.inodes[i]
+
+		key := make([]byte, len(inode.key))
+		copy(key, inode.key)
+		inode.key = key
+		_assert(len(inode.key) > 0, "dereference: zero-length inode key")
+
+		value := make([]byte, len(inode.value))
+		copy(value, inode.value)
+		inode.value = value
+	}
+
+	// Recursively dereference children.
+	for _, child := range n.children {
+		child.dereference()
+	}
+
+	// Update statistics.
+	n.bucket.tx.stats.NodeDeref++
+}
+
+// free adds the node's underlying page to the freelist.
+func (n *node) free() {
+	if n.pgid != 0 {
+		n.bucket.tx.db.freelist.free(n.bucket.tx.meta.txid, n.bucket.tx.page(n.pgid))
+		n.pgid = 0
+	}
+}
+
+// dump writes the contents of the node to STDERR for debugging purposes.
+/*
+func (n *node) dump() {
+	// Write node header.
+	var typ = "branch"
+	if n.isLeaf {
+		typ = "leaf"
+	}
+	warnf("[NODE %d {type=%s count=%d}]", n.pgid, typ, len(n.inodes))
+
+	// Write out abbreviated version of each item.
+	for _, item := range n.inodes {
+		if n.isLeaf {
+			if item.flags&bucketLeafFlag != 0 {
+				bucket := (*bucket)(unsafe.Pointer(&item.value[0]))
+				warnf("+L %08x -> (bucket root=%d)", trunc(item.key, 4), bucket.root)
+			} else {
+				warnf("+L %08x -> %08x", trunc(item.key, 4), trunc(item.value, 4))
+			}
+		} else {
+			warnf("+B %08x -> pgid=%d", trunc(item.key, 4), item.pgid)
+		}
+	}
+	warn("")
+}
+*/
+
+type nodes []*node
+
+func (s nodes) Len() int           { return len(s) }
+func (s nodes) Swap(i, j int)      { s[i], s[j] = s[j], s[i] }
+func (s nodes) Less(i, j int) bool { return bytes.Compare(s[i].inodes[0].key, s[j].inodes[0].key) == -1 }
+
+// inode represents an internal node inside of a node.
+// It can be used to point to elements in a page or point
+// to an element which hasn't been added to a page yet.
+type inode struct {
+	flags uint32
+	pgid  pgid
+	key   []byte
+	value []byte
+}
+
+type inodes []inode

vendor/github.com/coreos/bbolt/page.go

@@ -0,0 +1,197 @@
+package bolt
+
+import (
+	"fmt"
+	"os"
+	"sort"
+	"unsafe"
+)
+
+const pageHeaderSize = int(unsafe.Offsetof(((*page)(nil)).ptr))
+
+const minKeysPerPage = 2
+
+const branchPageElementSize = int(unsafe.Sizeof(branchPageElement{}))
+const leafPageElementSize = int(unsafe.Sizeof(leafPageElement{}))
+
+const (
+	branchPageFlag   = 0x01
+	leafPageFlag     = 0x02
+	metaPageFlag     = 0x04
+	freelistPageFlag = 0x10
+)
+
+const (
+	bucketLeafFlag = 0x01
+)
+
+type pgid uint64
+
+type page struct {
+	id       pgid
+	flags    uint16
+	count    uint16
+	overflow uint32
+	ptr      uintptr
+}
+
+// typ returns a human readable page type string used for debugging.
+func (p *page) typ() string {
+	if (p.flags & branchPageFlag) != 0 {
+		return "branch"
+	} else if (p.flags & leafPageFlag) != 0 {
+		return "leaf"
+	} else if (p.flags & metaPageFlag) != 0 {
+		return "meta"
+	} else if (p.flags & freelistPageFlag) != 0 {
+		return "freelist"
+	}
+	return fmt.Sprintf("unknown<%02x>", p.flags)
+}
+
+// meta returns a pointer to the metadata section of the page.
+func (p *page) meta() *meta {
+	return (*meta)(unsafe.Pointer(&p.ptr))
+}
+
+// leafPageElement retrieves the leaf node by index
+func (p *page) leafPageElement(index uint16) *leafPageElement {
+	n := &((*[0x7FFFFFF]leafPageElement)(unsafe.Pointer(&p.ptr)))[index]
+	return n
+}
+
+// leafPageElements retrieves a list of leaf nodes.
+func (p *page) leafPageElements() []leafPageElement {
+	if p.count == 0 {
+		return nil
+	}
+	return ((*[0x7FFFFFF]leafPageElement)(unsafe.Pointer(&p.ptr)))[:]
+}
+
+// branchPageElement retrieves the branch node by index
+func (p *page) branchPageElement(index uint16) *branchPageElement {
+	return &((*[0x7FFFFFF]branchPageElement)(unsafe.Pointer(&p.ptr)))[index]
+}
+
+// branchPageElements retrieves a list of branch nodes.
+func (p *page) branchPageElements() []branchPageElement {
+	if p.count == 0 {
+		return nil
+	}
+	return ((*[0x7FFFFFF]branchPageElement)(unsafe.Pointer(&p.ptr)))[:]
+}
+
+// dump writes n bytes of the page to STDERR as hex output.
+func (p *page) hexdump(n int) {
+	buf := (*[maxAllocSize]byte)(unsafe.Pointer(p))[:n]
+	fmt.Fprintf(os.Stderr, "%x\n", buf)
+}
+
+type pages []*page
+
+func (s pages) Len() int           { return len(s) }
+func (s pages) Swap(i, j int)      { s[i], s[j] = s[j], s[i] }
+func (s pages) Less(i, j int) bool { return s[i].id < s[j].id }
+
+// branchPageElement represents a node on a branch page.
+type branchPageElement struct {
+	pos   uint32
+	ksize uint32
+	pgid  pgid
+}
+
+// key returns a byte slice of the node key.
+func (n *branchPageElement) key() []byte {
+	buf := (*[maxAllocSize]byte)(unsafe.Pointer(n))
+	return (*[maxAllocSize]byte)(unsafe.Pointer(&buf[n.pos]))[:n.ksize]
+}
+
+// leafPageElement represents a node on a leaf page.
+type leafPageElement struct {
+	flags uint32
+	pos   uint32
+	ksize uint32
+	vsize uint32
+}
+
+// key returns a byte slice of the node key.
+func (n *leafPageElement) key() []byte {
+	buf := (*[maxAllocSize]byte)(unsafe.Pointer(n))
+	return (*[maxAllocSize]byte)(unsafe.Pointer(&buf[n.pos]))[:n.ksize:n.ksize]
+}
+
+// value returns a byte slice of the node value.
+func (n *leafPageElement) value() []byte {
+	buf := (*[maxAllocSize]byte)(unsafe.Pointer(n))
+	return (*[maxAllocSize]byte)(unsafe.Pointer(&buf[n.pos+n.ksize]))[:n.vsize:n.vsize]
+}
+
+// PageInfo represents human readable information about a page.
+type PageInfo struct {
+	ID            int
+	Type          string
+	Count         int
+	OverflowCount int
+}
+
+type pgids []pgid
+
+func (s pgids) Len() int           { return len(s) }
+func (s pgids) Swap(i, j int)      { s[i], s[j] = s[j], s[i] }
+func (s pgids) Less(i, j int) bool { return s[i] < s[j] }
+
+// merge returns the sorted union of a and b.
+func (a pgids) merge(b pgids) pgids {
+	// Return the opposite slice if one is nil.
+	if len(a) == 0 {
+		return b
+	}
+	if len(b) == 0 {
+		return a
+	}
+	merged := make(pgids, len(a)+len(b))
+	mergepgids(merged, a, b)
+	return merged
+}
+
+// mergepgids copies the sorted union of a and b into dst.
+// If dst is too small, it panics.
+func mergepgids(dst, a, b pgids) {
+	if len(dst) < len(a)+len(b) {
+		panic(fmt.Errorf("mergepgids bad len %d < %d + %d", len(dst), len(a), len(b)))
+	}
+	// Copy in the opposite slice if one is nil.
+	if len(a) == 0 {
+		copy(dst, b)
+		return
+	}
+	if len(b) == 0 {
+		copy(dst, a)
+		return
+	}
+
+	// Merged will hold all elements from both lists.
+	merged := dst[:0]
+
+	// Assign lead to the slice with a lower starting value, follow to the higher value.
+	lead, follow := a, b
+	if b[0] < a[0] {
+		lead, follow = b, a
+	}
+
+	// Continue while there are elements in the lead.
+	for len(lead) > 0 {
+		// Merge largest prefix of lead that is ahead of follow[0].
+		n := sort.Search(len(lead), func(i int) bool { return lead[i] > follow[0] })
+		merged = append(merged, lead[:n]...)
+		if n >= len(lead) {
+			break
+		}
+
+		// Swap lead and follow.
+		lead, follow = follow, lead[n:]
+	}
+
+	// Append what's left in follow.
+	_ = append(merged, follow...)
+}

vendor/github.com/coreos/bbolt/tx.go

@@ -0,0 +1,705 @@
+package bolt
+
+import (
+	"fmt"
+	"io"
+	"os"
+	"sort"
+	"strings"
+	"time"
+	"unsafe"
+)
+
+// txid represents the internal transaction identifier.
+type txid uint64
+
+// Tx represents a read-only or read/write transaction on the database.
+// Read-only transactions can be used for retrieving values for keys and creating cursors.
+// Read/write transactions can create and remove buckets and create and remove keys.
+//
+// IMPORTANT: You must commit or rollback transactions when you are done with
+// them. Pages can not be reclaimed by the writer until no more transactions
+// are using them. A long running read transaction can cause the database to
+// quickly grow.
+type Tx struct {
+	writable       bool
+	managed        bool
+	db             *DB
+	meta           *meta
+	root           Bucket
+	pages          map[pgid]*page
+	stats          TxStats
+	commitHandlers []func()
+
+	// WriteFlag specifies the flag for write-related methods like WriteTo().
+	// Tx opens the database file with the specified flag to copy the data.
+	//
+	// By default, the flag is unset, which works well for mostly in-memory
+	// workloads. For databases that are much larger than available RAM,
+	// set the flag to syscall.O_DIRECT to avoid trashing the page cache.
+	WriteFlag int
+}
+
+// init initializes the transaction.
+func (tx *Tx) init(db *DB) {
+	tx.db = db
+	tx.pages = nil
+
+	// Copy the meta page since it can be changed by the writer.
+	tx.meta = &meta{}
+	db.meta().copy(tx.meta)
+
+	// Copy over the root bucket.
+	tx.root = newBucket(tx)
+	tx.root.bucket = &bucket{}
+	*tx.root.bucket = tx.meta.root
+
+	// Increment the transaction id and add a page cache for writable transactions.
+	if tx.writable {
+		tx.pages = make(map[pgid]*page)
+		tx.meta.txid += txid(1)
+	}
+}
+
+// ID returns the transaction id.
+func (tx *Tx) ID() int {
+	return int(tx.meta.txid)
+}
+
+// DB returns a reference to the database that created the transaction.
+func (tx *Tx) DB() *DB {
+	return tx.db
+}
+
+// Size returns current database size in bytes as seen by this transaction.
+func (tx *Tx) Size() int64 {
+	return int64(tx.meta.pgid) * int64(tx.db.pageSize)
+}
+
+// Writable returns whether the transaction can perform write operations.
+func (tx *Tx) Writable() bool {
+	return tx.writable
+}
+
+// Cursor creates a cursor associated with the root bucket.
+// All items in the cursor will return a nil value because all root bucket keys point to buckets.
+// The cursor is only valid as long as the transaction is open.
+// Do not use a cursor after the transaction is closed.
+func (tx *Tx) Cursor() *Cursor {
+	return tx.root.Cursor()
+}
+
+// Stats retrieves a copy of the current transaction statistics.
+func (tx *Tx) Stats() TxStats {
+	return tx.stats
+}
+
+// Bucket retrieves a bucket by name.
+// Returns nil if the bucket does not exist.
+// The bucket instance is only valid for the lifetime of the transaction.
+func (tx *Tx) Bucket(name []byte) *Bucket {
+	return tx.root.Bucket(name)
+}
+
+// CreateBucket creates a new bucket.
+// Returns an error if the bucket already exists, if the bucket name is blank, or if the bucket name is too long.
+// The bucket instance is only valid for the lifetime of the transaction.
+func (tx *Tx) CreateBucket(name []byte) (*Bucket, error) {
+	return tx.root.CreateBucket(name)
+}
+
+// CreateBucketIfNotExists creates a new bucket if it doesn't already exist.
+// Returns an error if the bucket name is blank, or if the bucket name is too long.
+// The bucket instance is only valid for the lifetime of the transaction.
+func (tx *Tx) CreateBucketIfNotExists(name []byte) (*Bucket, error) {
+	return tx.root.CreateBucketIfNotExists(name)
+}
+
+// DeleteBucket deletes a bucket.
+// Returns an error if the bucket cannot be found or if the key represents a non-bucket value.
+func (tx *Tx) DeleteBucket(name []byte) error {
+	return tx.root.DeleteBucket(name)
+}
+
+// ForEach executes a function for each bucket in the root.
+// If the provided function returns an error then the iteration is stopped and
+// the error is returned to the caller.
+func (tx *Tx) ForEach(fn func(name []byte, b *Bucket) error) error {
+	return tx.root.ForEach(func(k, v []byte) error {
+		return fn(k, tx.root.Bucket(k))
+	})
+}
+
+// OnCommit adds a handler function to be executed after the transaction successfully commits.
+func (tx *Tx) OnCommit(fn func()) {
+	tx.commitHandlers = append(tx.commitHandlers, fn)
+}
+
+// Commit writes all changes to disk and updates the meta page.
+// Returns an error if a disk write error occurs, or if Commit is
+// called on a read-only transaction.
+func (tx *Tx) Commit() error {
+	_assert(!tx.managed, "managed tx commit not allowed")
+	if tx.db == nil {
+		return ErrTxClosed
+	} else if !tx.writable {
+		return ErrTxNotWritable
+	}
+
+	// TODO(benbjohnson): Use vectorized I/O to write out dirty pages.
+
+	// Rebalance nodes which have had deletions.
+	var startTime = time.Now()
+	tx.root.rebalance()
+	if tx.stats.Rebalance > 0 {
+		tx.stats.RebalanceTime += time.Since(startTime)
+	}
+
+	// spill data onto dirty pages.
+	startTime = time.Now()
+	if err := tx.root.spill(); err != nil {
+		tx.rollback()
+		return err
+	}
+	tx.stats.SpillTime += time.Since(startTime)
+
+	// Free the old root bucket.
+	tx.meta.root.root = tx.root.root
+
+	// Free the old freelist because commit writes out a fresh freelist.
+	if tx.meta.freelist != pgidNoFreelist {
+		tx.db.freelist.free(tx.meta.txid, tx.db.page(tx.meta.freelist))
+	}
+
+	if !tx.db.NoFreelistSync {
+		err := tx.commitFreelist()
+		if err != nil {
+			return err
+		}
+	} else {
+		tx.meta.freelist = pgidNoFreelist
+	}
+
+	// Write dirty pages to disk.
+	startTime = time.Now()
+	if err := tx.write(); err != nil {
+		tx.rollback()
+		return err
+	}
+
+	// If strict mode is enabled then perform a consistency check.
+	// Only the first consistency error is reported in the panic.
+	if tx.db.StrictMode {
+		ch := tx.Check()
+		var errs []string
+		for {
+			err, ok := <-ch
+			if !ok {
+				break
+			}
+			errs = append(errs, err.Error())
+		}
+		if len(errs) > 0 {
+			panic("check fail: " + strings.Join(errs, "\n"))
+		}
+	}
+
+	// Write meta to disk.
+	if err := tx.writeMeta(); err != nil {
+		tx.rollback()
+		return err
+	}
+	tx.stats.WriteTime += time.Since(startTime)
+
+	// Finalize the transaction.
+	tx.close()
+
+	// Execute commit handlers now that the locks have been removed.
+	for _, fn := range tx.commitHandlers {
+		fn()
+	}
+
+	return nil
+}
+
+func (tx *Tx) commitFreelist() error {
+	// Allocate new pages for the new free list. This will overestimate
+	// the size of the freelist but not underestimate the size (which would be bad).
+	opgid := tx.meta.pgid
+	p, err := tx.allocate((tx.db.freelist.size() / tx.db.pageSize) + 1)
+	if err != nil {
+		tx.rollback()
+		return err
+	}
+	if err := tx.db.freelist.write(p); err != nil {
+		tx.rollback()
+		return err
+	}
+	tx.meta.freelist = p.id
+	// If the high water mark has moved up then attempt to grow the database.
+	if tx.meta.pgid > opgid {
+		if err := tx.db.grow(int(tx.meta.pgid+1) * tx.db.pageSize); err != nil {
+			tx.rollback()
+			return err
+		}
+	}
+
+	return nil
+}
+
+// Rollback closes the transaction and ignores all previous updates. Read-only
+// transactions must be rolled back and not committed.
+func (tx *Tx) Rollback() error {
+	_assert(!tx.managed, "managed tx rollback not allowed")
+	if tx.db == nil {
+		return ErrTxClosed
+	}
+	tx.rollback()
+	return nil
+}
+
+func (tx *Tx) rollback() {
+	if tx.db == nil {
+		return
+	}
+	if tx.writable {
+		tx.db.freelist.rollback(tx.meta.txid)
+		tx.db.freelist.reload(tx.db.page(tx.db.meta().freelist))
+	}
+	tx.close()
+}
+
+func (tx *Tx) close() {
+	if tx.db == nil {
+		return
+	}
+	if tx.writable {
+		// Grab freelist stats.
+		var freelistFreeN = tx.db.freelist.free_count()
+		var freelistPendingN = tx.db.freelist.pending_count()
+		var freelistAlloc = tx.db.freelist.size()
+
+		// Remove transaction ref & writer lock.
+		tx.db.rwtx = nil
+		tx.db.rwlock.Unlock()
+
+		// Merge statistics.
+		tx.db.statlock.Lock()
+		tx.db.stats.FreePageN = freelistFreeN
+		tx.db.stats.PendingPageN = freelistPendingN
+		tx.db.stats.FreeAlloc = (freelistFreeN + freelistPendingN) * tx.db.pageSize
+		tx.db.stats.FreelistInuse = freelistAlloc
+		tx.db.stats.TxStats.add(&tx.stats)
+		tx.db.statlock.Unlock()
+	} else {
+		tx.db.removeTx(tx)
+	}
+
+	// Clear all references.
+	tx.db = nil
+	tx.meta = nil
+	tx.root = Bucket{tx: tx}
+	tx.pages = nil
+}
+
+// Copy writes the entire database to a writer.
+// This function exists for backwards compatibility. Use WriteTo() instead.
+func (tx *Tx) Copy(w io.Writer) error {
+	_, err := tx.WriteTo(w)
+	return err
+}
+
+// WriteTo writes the entire database to a writer.
+// If err == nil then exactly tx.Size() bytes will be written into the writer.
+func (tx *Tx) WriteTo(w io.Writer) (n int64, err error) {
+	// Attempt to open reader with WriteFlag
+	f, err := os.OpenFile(tx.db.path, os.O_RDONLY|tx.WriteFlag, 0)
+	if err != nil {
+		return 0, err
+	}
+	defer func() {
+		if cerr := f.Close(); err == nil {
+			err = cerr
+		}
+	}()
+
+	// Generate a meta page. We use the same page data for both meta pages.
+	buf := make([]byte, tx.db.pageSize)
+	page := (*page)(unsafe.Pointer(&buf[0]))
+	page.flags = metaPageFlag
+	*page.meta() = *tx.meta
+
+	// Write meta 0.
+	page.id = 0
+	page.meta().checksum = page.meta().sum64()
+	nn, err := w.Write(buf)
+	n += int64(nn)
+	if err != nil {
+		return n, fmt.Errorf("meta 0 copy: %s", err)
+	}
+
+	// Write meta 1 with a lower transaction id.
+	page.id = 1
+	page.meta().txid -= 1
+	page.meta().checksum = page.meta().sum64()
+	nn, err = w.Write(buf)
+	n += int64(nn)
+	if err != nil {
+		return n, fmt.Errorf("meta 1 copy: %s", err)
+	}
+
+	// Move past the meta pages in the file.
+	if _, err := f.Seek(int64(tx.db.pageSize*2), io.SeekStart); err != nil {
+		return n, fmt.Errorf("seek: %s", err)
+	}
+
+	// Copy data pages.
+	wn, err := io.CopyN(w, f, tx.Size()-int64(tx.db.pageSize*2))
+	n += wn
+	if err != nil {
+		return n, err
+	}
+
+	return n, nil
+}
+
+// CopyFile copies the entire database to file at the given path.
+// A reader transaction is maintained during the copy so it is safe to continue
+// using the database while a copy is in progress.
+func (tx *Tx) CopyFile(path string, mode os.FileMode) error {
+	f, err := os.OpenFile(path, os.O_RDWR|os.O_CREATE|os.O_TRUNC, mode)
+	if err != nil {
+		return err
+	}
+
+	err = tx.Copy(f)
+	if err != nil {
+		_ = f.Close()
+		return err
+	}
+	return f.Close()
+}
+
+// Check performs several consistency checks on the database for this transaction.
+// An error is returned if any inconsistency is found.
+//
+// It can be safely run concurrently on a writable transaction. However, this
+// incurs a high cost for large databases and databases with a lot of subbuckets
+// because of caching. This overhead can be removed if running on a read-only
+// transaction, however, it is not safe to execute other writer transactions at
+// the same time.
+func (tx *Tx) Check() <-chan error {
+	ch := make(chan error)
+	go tx.check(ch)
+	return ch
+}
+
+func (tx *Tx) check(ch chan error) {
+	// Force loading free list if opened in ReadOnly mode.
+	tx.db.loadFreelist()
+
+	// Check if any pages are double freed.
+	freed := make(map[pgid]bool)
+	all := make([]pgid, tx.db.freelist.count())
+	tx.db.freelist.copyall(all)
+	for _, id := range all {
+		if freed[id] {
+			ch <- fmt.Errorf("page %d: already freed", id)
+		}
+		freed[id] = true
+	}
+
+	// Track every reachable page.
+	reachable := make(map[pgid]*page)
+	reachable[0] = tx.page(0) // meta0
+	reachable[1] = tx.page(1) // meta1
+	if tx.meta.freelist != pgidNoFreelist {
+		for i := uint32(0); i <= tx.page(tx.meta.freelist).overflow; i++ {
+			reachable[tx.meta.freelist+pgid(i)] = tx.page(tx.meta.freelist)
+		}
+	}
+
+	// Recursively check buckets.
+	tx.checkBucket(&tx.root, reachable, freed, ch)
+
+	// Ensure all pages below high water mark are either reachable or freed.
+	for i := pgid(0); i < tx.meta.pgid; i++ {
+		_, isReachable := reachable[i]
+		if !isReachable && !freed[i] {
+			ch <- fmt.Errorf("page %d: unreachable unfreed", int(i))
+		}
+	}
+
+	// Close the channel to signal completion.
+	close(ch)
+}
+
+func (tx *Tx) checkBucket(b *Bucket, reachable map[pgid]*page, freed map[pgid]bool, ch chan error) {
+	// Ignore inline buckets.
+	if b.root == 0 {
+		return
+	}
+
+	// Check every page used by this bucket.
+	b.tx.forEachPage(b.root, 0, func(p *page, _ int) {
+		if p.id > tx.meta.pgid {
+			ch <- fmt.Errorf("page %d: out of bounds: %d", int(p.id), int(b.tx.meta.pgid))
+		}
+
+		// Ensure each page is only referenced once.
+		for i := pgid(0); i <= pgid(p.overflow); i++ {
+			var id = p.id + i
+			if _, ok := reachable[id]; ok {
+				ch <- fmt.Errorf("page %d: multiple references", int(id))
+			}
+			reachable[id] = p
+		}
+
+		// We should only encounter un-freed leaf and branch pages.
+		if freed[p.id] {
+			ch <- fmt.Errorf("page %d: reachable freed", int(p.id))
+		} else if (p.flags&branchPageFlag) == 0 && (p.flags&leafPageFlag) == 0 {
+			ch <- fmt.Errorf("page %d: invalid type: %s", int(p.id), p.typ())
+		}
+	})
+
+	// Check each bucket within this bucket.
+	_ = b.ForEach(func(k, v []byte) error {
+		if child := b.Bucket(k); child != nil {
+			tx.checkBucket(child, reachable, freed, ch)
+		}
+		return nil
+	})
+}
+
+// allocate returns a contiguous block of memory starting at a given page.
+func (tx *Tx) allocate(count int) (*page, error) {
+	p, err := tx.db.allocate(tx.meta.txid, count)
+	if err != nil {
+		return nil, err
+	}
+
+	// Save to our page cache.
+	tx.pages[p.id] = p
+
+	// Update statistics.
+	tx.stats.PageCount++
+	tx.stats.PageAlloc += count * tx.db.pageSize
+
+	return p, nil
+}
+
+// write writes any dirty pages to disk.
+func (tx *Tx) write() error {
+	// Sort pages by id.
+	pages := make(pages, 0, len(tx.pages))
+	for _, p := range tx.pages {
+		pages = append(pages, p)
+	}
+	// Clear out page cache early.
+	tx.pages = make(map[pgid]*page)
+	sort.Sort(pages)
+
+	// Write pages to disk in order.
+	for _, p := range pages {
+		size := (int(p.overflow) + 1) * tx.db.pageSize
+		offset := int64(p.id) * int64(tx.db.pageSize)
+
+		// Write out page in "max allocation" sized chunks.
+		ptr := (*[maxAllocSize]byte)(unsafe.Pointer(p))
+		for {
+			// Limit our write to our max allocation size.
+			sz := size
+			if sz > maxAllocSize-1 {
+				sz = maxAllocSize - 1
+			}
+
+			// Write chunk to disk.
+			buf := ptr[:sz]
+			if _, err := tx.db.ops.writeAt(buf, offset); err != nil {
+				return err
+			}
+
+			// Update statistics.
+			tx.stats.Write++
+
+			// Exit inner for loop if we've written all the chunks.
+			size -= sz
+			if size == 0 {
+				break
+			}
+
+			// Otherwise move offset forward and move pointer to next chunk.
+			offset += int64(sz)
+			ptr = (*[maxAllocSize]byte)(unsafe.Pointer(&ptr[sz]))
+		}
+	}
+
+	// Ignore file sync if flag is set on DB.
+	if !tx.db.NoSync || IgnoreNoSync {
+		if err := fdatasync(tx.db); err != nil {
+			return err
+		}
+	}
+
+	// Put small pages back to page pool.
+	for _, p := range pages {
+		// Ignore page sizes over 1 page.
+		// These are allocated using make() instead of the page pool.
+		if int(p.overflow) != 0 {
+			continue
+		}
+
+		buf := (*[maxAllocSize]byte)(unsafe.Pointer(p))[:tx.db.pageSize]
+
+		// See https://go.googlesource.com/go/+/f03c9202c43e0abb130669852082117ca50aa9b1
+		for i := range buf {
+			buf[i] = 0
+		}
+		tx.db.pagePool.Put(buf)
+	}
+
+	return nil
+}
+
+// writeMeta writes the meta to the disk.
+func (tx *Tx) writeMeta() error {
+	// Create a temporary buffer for the meta page.
+	buf := make([]byte, tx.db.pageSize)
+	p := tx.db.pageInBuffer(buf, 0)
+	tx.meta.write(p)
+
+	// Write the meta page to file.
+	if _, err := tx.db.ops.writeAt(buf, int64(p.id)*int64(tx.db.pageSize)); err != nil {
+		return err
+	}
+	if !tx.db.NoSync || IgnoreNoSync {
+		if err := fdatasync(tx.db); err != nil {
+			return err
+		}
+	}
+
+	// Update statistics.
+	tx.stats.Write++
+
+	return nil
+}
+
+// page returns a reference to the page with a given id.
+// If page has been written to then a temporary buffered page is returned.
+func (tx *Tx) page(id pgid) *page {
+	// Check the dirty pages first.
+	if tx.pages != nil {
+		if p, ok := tx.pages[id]; ok {
+			return p
+		}
+	}
+
+	// Otherwise return directly from the mmap.
+	return tx.db.page(id)
+}
+
+// forEachPage iterates over every page within a given page and executes a function.
+func (tx *Tx) forEachPage(pgid pgid, depth int, fn func(*page, int)) {
+	p := tx.page(pgid)
+
+	// Execute function.
+	fn(p, depth)
+
+	// Recursively loop over children.
+	if (p.flags & branchPageFlag) != 0 {
+		for i := 0; i < int(p.count); i++ {
+			elem := p.branchPageElement(uint16(i))
+			tx.forEachPage(elem.pgid, depth+1, fn)
+		}
+	}
+}
+
+// Page returns page information for a given page number.
+// This is only safe for concurrent use when used by a writable transaction.
+func (tx *Tx) Page(id int) (*PageInfo, error) {
+	if tx.db == nil {
+		return nil, ErrTxClosed
+	} else if pgid(id) >= tx.meta.pgid {
+		return nil, nil
+	}
+
+	// Build the page info.
+	p := tx.db.page(pgid(id))
+	info := &PageInfo{
+		ID:            id,
+		Count:         int(p.count),
+		OverflowCount: int(p.overflow),
+	}
+
+	// Determine the type (or if it's free).
+	if tx.db.freelist.freed(pgid(id)) {
+		info.Type = "free"
+	} else {
+		info.Type = p.typ()
+	}
+
+	return info, nil
+}
+
+// TxStats represents statistics about the actions performed by the transaction.
+type TxStats struct {
+	// Page statistics.
+	PageCount int // number of page allocations
+	PageAlloc int // total bytes allocated
+
+	// Cursor statistics.
+	CursorCount int // number of cursors created
+
+	// Node statistics
+	NodeCount int // number of node allocations
+	NodeDeref int // number of node dereferences
+
+	// Rebalance statistics.
+	Rebalance     int           // number of node rebalances
+	RebalanceTime time.Duration // total time spent rebalancing
+
+	// Split/Spill statistics.
+	Split     int           // number of nodes split
+	Spill     int           // number of nodes spilled
+	SpillTime time.Duration // total time spent spilling
+
+	// Write statistics.
+	Write     int           // number of writes performed
+	WriteTime time.Duration // total time spent writing to disk
+}
+
+func (s *TxStats) add(other *TxStats) {
+	s.PageCount += other.PageCount
+	s.PageAlloc += other.PageAlloc
+	s.CursorCount += other.CursorCount
+	s.NodeCount += other.NodeCount
+	s.NodeDeref += other.NodeDeref
+	s.Rebalance += other.Rebalance
+	s.RebalanceTime += other.RebalanceTime
+	s.Split += other.Split
+	s.Spill += other.Spill
+	s.SpillTime += other.SpillTime
+	s.Write += other.Write
+	s.WriteTime += other.WriteTime
+}
+
+// Sub calculates and returns the difference between two sets of transaction stats.
+// This is useful when obtaining stats at two different points and time and
+// you need the performance counters that occurred within that time span.
+func (s *TxStats) Sub(other *TxStats) TxStats {
+	var diff TxStats
+	diff.PageCount = s.PageCount - other.PageCount
+	diff.PageAlloc = s.PageAlloc - other.PageAlloc
+	diff.CursorCount = s.CursorCount - other.CursorCount
+	diff.NodeCount = s.NodeCount - other.NodeCount
+	diff.NodeDeref = s.NodeDeref - other.NodeDeref
+	diff.Rebalance = s.Rebalance - other.Rebalance
+	diff.RebalanceTime = s.RebalanceTime - other.RebalanceTime
+	diff.Split = s.Split - other.Split
+	diff.Spill = s.Spill - other.Spill
+	diff.SpillTime = s.SpillTime - other.SpillTime
+	diff.Write = s.Write - other.Write
+	diff.WriteTime = s.WriteTime - other.WriteTime
+	return diff
+}

vendor/github.com/coreos/etcd/auth/authpb/auth.pb.go

@@ -0,0 +1,824 @@
+// Code generated by protoc-gen-gogo.
+// source: auth.proto
+// DO NOT EDIT!
+
+/*
+	Package authpb is a generated protocol buffer package.
+
+	It is generated from these files:
+		auth.proto
+
+	It has these top-level messages:
+		User
+		Permission
+		Role
+*/
+package authpb
+
+import (
+	"fmt"
+
+	proto "github.com/golang/protobuf/proto"
+
+	math "math"
+
+	io "io"
+)
+
+// Reference imports to suppress errors if they are not otherwise used.
+var _ = proto.Marshal
+var _ = fmt.Errorf
+var _ = math.Inf
+
+// This is a compile-time assertion to ensure that this generated file
+// is compatible with the proto package it is being compiled against.
+// A compilation error at this line likely means your copy of the
+// proto package needs to be updated.
+const _ = proto.ProtoPackageIsVersion2 // please upgrade the proto package
+
+type Permission_Type int32
+
+const (
+	READ      Permission_Type = 0
+	WRITE     Permission_Type = 1
+	READWRITE Permission_Type = 2
+)
+
+var Permission_Type_name = map[int32]string{
+	0: "READ",
+	1: "WRITE",
+	2: "READWRITE",
+}
+var Permission_Type_value = map[string]int32{
+	"READ":      0,
+	"WRITE":     1,
+	"READWRITE": 2,
+}
+
+func (x Permission_Type) String() string {
+	return proto.EnumName(Permission_Type_name, int32(x))
+}
+func (Permission_Type) EnumDescriptor() ([]byte, []int) { return fileDescriptorAuth, []int{1, 0} }
+
+// User is a single entry in the bucket authUsers
+type User struct {
+	Name     []byte   `protobuf:"bytes,1,opt,name=name,proto3" json:"name,omitempty"`
+	Password []byte   `protobuf:"bytes,2,opt,name=password,proto3" json:"password,omitempty"`
+	Roles    []string `protobuf:"bytes,3,rep,name=roles" json:"roles,omitempty"`
+}
+
+func (m *User) Reset()                    { *m = User{} }
+func (m *User) String() string            { return proto.CompactTextString(m) }
+func (*User) ProtoMessage()               {}
+func (*User) Descriptor() ([]byte, []int) { return fileDescriptorAuth, []int{0} }
+
+// Permission is a single entity
+type Permission struct {
+	PermType Permission_Type `protobuf:"varint,1,opt,name=permType,proto3,enum=authpb.Permission_Type" json:"permType,omitempty"`
+	Key      []byte          `protobuf:"bytes,2,opt,name=key,proto3" json:"key,omitempty"`
+	RangeEnd []byte          `protobuf:"bytes,3,opt,name=range_end,json=rangeEnd,proto3" json:"range_end,omitempty"`
+}
+
+func (m *Permission) Reset()                    { *m = Permission{} }
+func (m *Permission) String() string            { return proto.CompactTextString(m) }
+func (*Permission) ProtoMessage()               {}
+func (*Permission) Descriptor() ([]byte, []int) { return fileDescriptorAuth, []int{1} }
+
+// Role is a single entry in the bucket authRoles
+type Role struct {
+	Name          []byte        `protobuf:"bytes,1,opt,name=name,proto3" json:"name,omitempty"`
+	KeyPermission []*Permission `protobuf:"bytes,2,rep,name=keyPermission" json:"keyPermission,omitempty"`
+}
+
+func (m *Role) Reset()                    { *m = Role{} }
+func (m *Role) String() string            { return proto.CompactTextString(m) }
+func (*Role) ProtoMessage()               {}
+func (*Role) Descriptor() ([]byte, []int) { return fileDescriptorAuth, []int{2} }
+
+func init() {
+	proto.RegisterType((*User)(nil), "authpb.User")
+	proto.RegisterType((*Permission)(nil), "authpb.Permission")
+	proto.RegisterType((*Role)(nil), "authpb.Role")
+	proto.RegisterEnum("authpb.Permission_Type", Permission_Type_name, Permission_Type_value)
+}
+func (m *User) Marshal() (dAtA []byte, err error) {
+	size := m.Size()
+	dAtA = make([]byte, size)
+	n, err := m.MarshalTo(dAtA)
+	if err != nil {
+		return nil, err
+	}
+	return dAtA[:n], nil
+}
+
+func (m *User) MarshalTo(dAtA []byte) (int, error) {
+	var i int
+	_ = i
+	var l int
+	_ = l
+	if len(m.Name) > 0 {
+		dAtA[i] = 0xa
+		i++
+		i = encodeVarintAuth(dAtA, i, uint64(len(m.Name)))
+		i += copy(dAtA[i:], m.Name)
+	}
+	if len(m.Password) > 0 {
+		dAtA[i] = 0x12
+		i++
+		i = encodeVarintAuth(dAtA, i, uint64(len(m.Password)))
+		i += copy(dAtA[i:], m.Password)
+	}
+	if len(m.Roles) > 0 {
+		for _, s := range m.Roles {
+			dAtA[i] = 0x1a
+			i++
+			l = len(s)
+			for l >= 1<<7 {
+				dAtA[i] = uint8(uint64(l)&0x7f | 0x80)
+				l >>= 7
+				i++
+			}
+			dAtA[i] = uint8(l)
+			i++
+			i += copy(dAtA[i:], s)
+		}
+	}
+	return i, nil
+}
+
+func (m *Permission) Marshal() (dAtA []byte, err error) {
+	size := m.Size()
+	dAtA = make([]byte, size)
+	n, err := m.MarshalTo(dAtA)
+	if err != nil {
+		return nil, err
+	}
+	return dAtA[:n], nil
+}
+
+func (m *Permission) MarshalTo(dAtA []byte) (int, error) {
+	var i int
+	_ = i
+	var l int
+	_ = l
+	if m.PermType != 0 {
+		dAtA[i] = 0x8
+		i++
+		i = encodeVarintAuth(dAtA, i, uint64(m.PermType))
+	}
+	if len(m.Key) > 0 {
+		dAtA[i] = 0x12
+		i++
+		i = encodeVarintAuth(dAtA, i, uint64(len(m.Key)))
+		i += copy(dAtA[i:], m.Key)
+	}
+	if len(m.RangeEnd) > 0 {
+		dAtA[i] = 0x1a
+		i++
+		i = encodeVarintAuth(dAtA, i, uint64(len(m.RangeEnd)))
+		i += copy(dAtA[i:], m.RangeEnd)
+	}
+	return i, nil
+}
+
+func (m *Role) Marshal() (dAtA []byte, err error) {
+	size := m.Size()
+	dAtA = make([]byte, size)
+	n, err := m.MarshalTo(dAtA)
+	if err != nil {
+		return nil, err
+	}
+	return dAtA[:n], nil
+}
+
+func (m *Role) MarshalTo(dAtA []byte) (int, error) {
+	var i int
+	_ = i
+	var l int
+	_ = l
+	if len(m.Name) > 0 {
+		dAtA[i] = 0xa
+		i++
+		i = encodeVarintAuth(dAtA, i, uint64(len(m.Name)))
+		i += copy(dAtA[i:], m.Name)
+	}
+	if len(m.KeyPermission) > 0 {
+		for _, msg := range m.KeyPermission {
+			dAtA[i] = 0x12
+			i++
+			i = encodeVarintAuth(dAtA, i, uint64(msg.Size()))
+			n, err := msg.MarshalTo(dAtA[i:])
+			if err != nil {
+				return 0, err
+			}
+			i += n
+		}
+	}
+	return i, nil
+}
+
+func encodeFixed64Auth(dAtA []byte, offset int, v uint64) int {
+	dAtA[offset] = uint8(v)
+	dAtA[offset+1] = uint8(v >> 8)
+	dAtA[offset+2] = uint8(v >> 16)
+	dAtA[offset+3] = uint8(v >> 24)
+	dAtA[offset+4] = uint8(v >> 32)
+	dAtA[offset+5] = uint8(v >> 40)
+	dAtA[offset+6] = uint8(v >> 48)
+	dAtA[offset+7] = uint8(v >> 56)
+	return offset + 8
+}
+func encodeFixed32Auth(dAtA []byte, offset int, v uint32) int {
+	dAtA[offset] = uint8(v)
+	dAtA[offset+1] = uint8(v >> 8)
+	dAtA[offset+2] = uint8(v >> 16)
+	dAtA[offset+3] = uint8(v >> 24)
+	return offset + 4
+}
+func encodeVarintAuth(dAtA []byte, offset int, v uint64) int {
+	for v >= 1<<7 {
+		dAtA[offset] = uint8(v&0x7f | 0x80)
+		v >>= 7
+		offset++
+	}
+	dAtA[offset] = uint8(v)
+	return offset + 1
+}
+func (m *User) Size() (n int) {
+	var l int
+	_ = l
+	l = len(m.Name)
+	if l > 0 {
+		n += 1 + l + sovAuth(uint64(l))
+	}
+	l = len(m.Password)
+	if l > 0 {
+		n += 1 + l + sovAuth(uint64(l))
+	}
+	if len(m.Roles) > 0 {
+		for _, s := range m.Roles {
+			l = len(s)
+			n += 1 + l + sovAuth(uint64(l))
+		}
+	}
+	return n
+}
+
+func (m *Permission) Size() (n int) {
+	var l int
+	_ = l
+	if m.PermType != 0 {
+		n += 1 + sovAuth(uint64(m.PermType))
+	}
+	l = len(m.Key)
+	if l > 0 {
+		n += 1 + l + sovAuth(uint64(l))
+	}
+	l = len(m.RangeEnd)
+	if l > 0 {
+		n += 1 + l + sovAuth(uint64(l))
+	}
+	return n
+}
+
+func (m *Role) Size() (n int) {
+	var l int
+	_ = l
+	l = len(m.Name)
+	if l > 0 {
+		n += 1 + l + sovAuth(uint64(l))
+	}
+	if len(m.KeyPermission) > 0 {
+		for _, e := range m.KeyPermission {
+			l = e.Size()
+			n += 1 + l + sovAuth(uint64(l))
+		}
+	}
+	return n
+}
+
+func sovAuth(x uint64) (n int) {
+	for {
+		n++
+		x >>= 7
+		if x == 0 {
+			break
+		}
+	}
+	return n
+}
+func sozAuth(x uint64) (n int) {
+	return sovAuth(uint64((x << 1) ^ uint64((int64(x) >> 63))))
+}
+func (m *User) Unmarshal(dAtA []byte) error {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		preIndex := iNdEx
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return ErrIntOverflowAuth
+			}
+			if iNdEx >= l {
+				return io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		fieldNum := int32(wire >> 3)
+		wireType := int(wire & 0x7)
+		if wireType == 4 {
+			return fmt.Errorf("proto: User: wiretype end group for non-group")
+		}
+		if fieldNum <= 0 {
+			return fmt.Errorf("proto: User: illegal tag %d (wire type %d)", fieldNum, wire)
+		}
+		switch fieldNum {
+		case 1:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Name", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Name = append(m.Name[:0], dAtA[iNdEx:postIndex]...)
+			if m.Name == nil {
+				m.Name = []byte{}
+			}
+			iNdEx = postIndex
+		case 2:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Password", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Password = append(m.Password[:0], dAtA[iNdEx:postIndex]...)
+			if m.Password == nil {
+				m.Password = []byte{}
+			}
+			iNdEx = postIndex
+		case 3:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Roles", wireType)
+			}
+			var stringLen uint64
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				stringLen |= (uint64(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			intStringLen := int(stringLen)
+			if intStringLen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + intStringLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Roles = append(m.Roles, string(dAtA[iNdEx:postIndex]))
+			iNdEx = postIndex
+		default:
+			iNdEx = preIndex
+			skippy, err := skipAuth(dAtA[iNdEx:])
+			if err != nil {
+				return err
+			}
+			if skippy < 0 {
+				return ErrInvalidLengthAuth
+			}
+			if (iNdEx + skippy) > l {
+				return io.ErrUnexpectedEOF
+			}
+			iNdEx += skippy
+		}
+	}
+
+	if iNdEx > l {
+		return io.ErrUnexpectedEOF
+	}
+	return nil
+}
+func (m *Permission) Unmarshal(dAtA []byte) error {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		preIndex := iNdEx
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return ErrIntOverflowAuth
+			}
+			if iNdEx >= l {
+				return io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		fieldNum := int32(wire >> 3)
+		wireType := int(wire & 0x7)
+		if wireType == 4 {
+			return fmt.Errorf("proto: Permission: wiretype end group for non-group")
+		}
+		if fieldNum <= 0 {
+			return fmt.Errorf("proto: Permission: illegal tag %d (wire type %d)", fieldNum, wire)
+		}
+		switch fieldNum {
+		case 1:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field PermType", wireType)
+			}
+			m.PermType = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.PermType |= (Permission_Type(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		case 2:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Key", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Key = append(m.Key[:0], dAtA[iNdEx:postIndex]...)
+			if m.Key == nil {
+				m.Key = []byte{}
+			}
+			iNdEx = postIndex
+		case 3:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field RangeEnd", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.RangeEnd = append(m.RangeEnd[:0], dAtA[iNdEx:postIndex]...)
+			if m.RangeEnd == nil {
+				m.RangeEnd = []byte{}
+			}
+			iNdEx = postIndex
+		default:
+			iNdEx = preIndex
+			skippy, err := skipAuth(dAtA[iNdEx:])
+			if err != nil {
+				return err
+			}
+			if skippy < 0 {
+				return ErrInvalidLengthAuth
+			}
+			if (iNdEx + skippy) > l {
+				return io.ErrUnexpectedEOF
+			}
+			iNdEx += skippy
+		}
+	}
+
+	if iNdEx > l {
+		return io.ErrUnexpectedEOF
+	}
+	return nil
+}
+func (m *Role) Unmarshal(dAtA []byte) error {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		preIndex := iNdEx
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return ErrIntOverflowAuth
+			}
+			if iNdEx >= l {
+				return io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		fieldNum := int32(wire >> 3)
+		wireType := int(wire & 0x7)
+		if wireType == 4 {
+			return fmt.Errorf("proto: Role: wiretype end group for non-group")
+		}
+		if fieldNum <= 0 {
+			return fmt.Errorf("proto: Role: illegal tag %d (wire type %d)", fieldNum, wire)
+		}
+		switch fieldNum {
+		case 1:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Name", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Name = append(m.Name[:0], dAtA[iNdEx:postIndex]...)
+			if m.Name == nil {
+				m.Name = []byte{}
+			}
+			iNdEx = postIndex
+		case 2:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field KeyPermission", wireType)
+			}
+			var msglen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				msglen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if msglen < 0 {
+				return ErrInvalidLengthAuth
+			}
+			postIndex := iNdEx + msglen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.KeyPermission = append(m.KeyPermission, &Permission{})
+			if err := m.KeyPermission[len(m.KeyPermission)-1].Unmarshal(dAtA[iNdEx:postIndex]); err != nil {
+				return err
+			}
+			iNdEx = postIndex
+		default:
+			iNdEx = preIndex
+			skippy, err := skipAuth(dAtA[iNdEx:])
+			if err != nil {
+				return err
+			}
+			if skippy < 0 {
+				return ErrInvalidLengthAuth
+			}
+			if (iNdEx + skippy) > l {
+				return io.ErrUnexpectedEOF
+			}
+			iNdEx += skippy
+		}
+	}
+
+	if iNdEx > l {
+		return io.ErrUnexpectedEOF
+	}
+	return nil
+}
+func skipAuth(dAtA []byte) (n int, err error) {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return 0, ErrIntOverflowAuth
+			}
+			if iNdEx >= l {
+				return 0, io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		wireType := int(wire & 0x7)
+		switch wireType {
+		case 0:
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return 0, ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return 0, io.ErrUnexpectedEOF
+				}
+				iNdEx++
+				if dAtA[iNdEx-1] < 0x80 {
+					break
+				}
+			}
+			return iNdEx, nil
+		case 1:
+			iNdEx += 8
+			return iNdEx, nil
+		case 2:
+			var length int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return 0, ErrIntOverflowAuth
+				}
+				if iNdEx >= l {
+					return 0, io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				length |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			iNdEx += length
+			if length < 0 {
+				return 0, ErrInvalidLengthAuth
+			}
+			return iNdEx, nil
+		case 3:
+			for {
+				var innerWire uint64
+				var start int = iNdEx
+				for shift := uint(0); ; shift += 7 {
+					if shift >= 64 {
+						return 0, ErrIntOverflowAuth
+					}
+					if iNdEx >= l {
+						return 0, io.ErrUnexpectedEOF
+					}
+					b := dAtA[iNdEx]
+					iNdEx++
+					innerWire |= (uint64(b) & 0x7F) << shift
+					if b < 0x80 {
+						break
+					}
+				}
+				innerWireType := int(innerWire & 0x7)
+				if innerWireType == 4 {
+					break
+				}
+				next, err := skipAuth(dAtA[start:])
+				if err != nil {
+					return 0, err
+				}
+				iNdEx = start + next
+			}
+			return iNdEx, nil
+		case 4:
+			return iNdEx, nil
+		case 5:
+			iNdEx += 4
+			return iNdEx, nil
+		default:
+			return 0, fmt.Errorf("proto: illegal wireType %d", wireType)
+		}
+	}
+	panic("unreachable")
+}
+
+var (
+	ErrInvalidLengthAuth = fmt.Errorf("proto: negative length found during unmarshaling")
+	ErrIntOverflowAuth   = fmt.Errorf("proto: integer overflow")
+)
+
+func init() { proto.RegisterFile("auth.proto", fileDescriptorAuth) }
+
+var fileDescriptorAuth = []byte{
+	// 288 bytes of a gzipped FileDescriptorProto
+	0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0x6c, 0x90, 0xc1, 0x4a, 0xc3, 0x30,
+	0x1c, 0xc6, 0x9b, 0xb6, 0x1b, 0xed, 0x5f, 0x27, 0x25, 0x0c, 0x0c, 0x13, 0x42, 0xe9, 0xa9, 0x78,
+	0xa8, 0xb0, 0x5d, 0xbc, 0x2a, 0xf6, 0x20, 0x78, 0x90, 0x50, 0xf1, 0x28, 0x1d, 0x0d, 0x75, 0x6c,
+	0x6d, 0x4a, 0x32, 0x91, 0xbe, 0x89, 0x07, 0x1f, 0x68, 0xc7, 0x3d, 0x82, 0xab, 0x2f, 0x22, 0x4d,
+	0x64, 0x43, 0xdc, 0xed, 0xfb, 0xbe, 0xff, 0x97, 0xe4, 0x97, 0x3f, 0x40, 0xfe, 0xb6, 0x7e, 0x4d,
+	0x1a, 0x29, 0xd6, 0x02, 0x0f, 0x7b, 0xdd, 0xcc, 0x27, 0xe3, 0x52, 0x94, 0x42, 0x47, 0x57, 0xbd,
+	0x32, 0xd3, 0xe8, 0x01, 0xdc, 0x27, 0xc5, 0x25, 0xc6, 0xe0, 0xd6, 0x79, 0xc5, 0x09, 0x0a, 0x51,
+	0x7c, 0xca, 0xb4, 0xc6, 0x13, 0xf0, 0x9a, 0x5c, 0xa9, 0x77, 0x21, 0x0b, 0x62, 0xeb, 0x7c, 0xef,
+	0xf1, 0x18, 0x06, 0x52, 0xac, 0xb8, 0x22, 0x4e, 0xe8, 0xc4, 0x3e, 0x33, 0x26, 0xfa, 0x44, 0x00,
+	0x8f, 0x5c, 0x56, 0x0b, 0xa5, 0x16, 0xa2, 0xc6, 0x33, 0xf0, 0x1a, 0x2e, 0xab, 0xac, 0x6d, 0xcc,
+	0xc5, 0x67, 0xd3, 0xf3, 0xc4, 0xd0, 0x24, 0x87, 0x56, 0xd2, 0x8f, 0xd9, 0xbe, 0x88, 0x03, 0x70,
+	0x96, 0xbc, 0xfd, 0x7d, 0xb0, 0x97, 0xf8, 0x02, 0x7c, 0x99, 0xd7, 0x25, 0x7f, 0xe1, 0x75, 0x41,
+	0x1c, 0x03, 0xa2, 0x83, 0xb4, 0x2e, 0xa2, 0x4b, 0x70, 0xf5, 0x31, 0x0f, 0x5c, 0x96, 0xde, 0xdc,
+	0x05, 0x16, 0xf6, 0x61, 0xf0, 0xcc, 0xee, 0xb3, 0x34, 0x40, 0x78, 0x04, 0x7e, 0x1f, 0x1a, 0x6b,
+	0x47, 0x19, 0xb8, 0x4c, 0xac, 0xf8, 0xd1, 0xcf, 0x5e, 0xc3, 0x68, 0xc9, 0xdb, 0x03, 0x16, 0xb1,
+	0x43, 0x27, 0x3e, 0x99, 0xe2, 0xff, 0xc0, 0xec, 0x6f, 0xf1, 0x96, 0x6c, 0x76, 0xd4, 0xda, 0xee,
+	0xa8, 0xb5, 0xe9, 0x28, 0xda, 0x76, 0x14, 0x7d, 0x75, 0x14, 0x7d, 0x7c, 0x53, 0x6b, 0x3e, 0xd4,
+	0x3b, 0x9e, 0xfd, 0x04, 0x00, 0x00, 0xff, 0xff, 0xcc, 0x76, 0x8d, 0x4f, 0x8f, 0x01, 0x00, 0x00,
+}

vendor/github.com/coreos/etcd/client/auth_role.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -56,22 +56,22 @@ func NewAuthRoleAPI(c Client) AuthRoleAPI {
 }
 
 type AuthRoleAPI interface {
-	// Add a role.
+	// AddRole adds a role.
 	AddRole(ctx context.Context, role string) error
 
-	// Remove a role.
+	// RemoveRole removes a role.
 	RemoveRole(ctx context.Context, role string) error
 
-	// Get role details.
+	// GetRole retrieves role details.
 	GetRole(ctx context.Context, role string) (*Role, error)
 
-	// Grant a role some permission prefixes for the KV store.
+	// GrantRoleKV grants a role some permission prefixes for the KV store.
 	GrantRoleKV(ctx context.Context, role string, prefixes []string, permType PermissionType) (*Role, error)
 
-	// Revoke some some permission prefixes for a role on the KV store.
+	// RevokeRoleKV revokes some permission prefixes for a role on the KV store.
 	RevokeRoleKV(ctx context.Context, role string, prefixes []string, permType PermissionType) (*Role, error)
 
-	// List roles.
+	// ListRoles lists roles.
 	ListRoles(ctx context.Context) ([]string, error)
 }
 
@@ -115,17 +115,20 @@ func (r *httpAuthRoleAPI) ListRoles(ctx context.Context) ([]string, error) {
 	if err != nil {
 		return nil, err
 	}
-	if err := assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
+	if err = assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
 		return nil, err
 	}
-	var userList struct {
+	var roleList struct {
-		Roles []string `json:"roles"`
+		Roles []Role `json:"roles"`
 	}
-	err = json.Unmarshal(body, &userList)
+	if err = json.Unmarshal(body, &roleList); err != nil {
-	if err != nil {
 		return nil, err
 	}
-	return userList.Roles, nil
+	ret := make([]string, 0, len(roleList.Roles))
+	for _, r := range roleList.Roles {
+		ret = append(ret, r.Role)
+	}
+	return ret, nil
 }
 
 func (r *httpAuthRoleAPI) AddRole(ctx context.Context, rolename string) error {
@@ -218,17 +221,16 @@ func (r *httpAuthRoleAPI) modRole(ctx context.Context, req *authRoleAPIAction) (
 	if err != nil {
 		return nil, err
 	}
-	if err := assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
+	if err = assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
 		var sec authError
-		err := json.Unmarshal(body, &sec)
+		err = json.Unmarshal(body, &sec)
 		if err != nil {
 			return nil, err
 		}
 		return nil, sec
 	}
 	var role Role
-	err = json.Unmarshal(body, &role)
+	if err = json.Unmarshal(body, &role); err != nil {
-	if err != nil {
 		return nil, err
 	}
 	return &role, nil

vendor/github.com/coreos/etcd/client/auth_user.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -36,6 +36,17 @@ type User struct {
 	Revoke   []string `json:"revoke,omitempty"`
 }
 
+// userListEntry is the user representation given by the server for ListUsers
+type userListEntry struct {
+	User  string `json:"user"`
+	Roles []Role `json:"roles"`
+}
+
+type UserRoles struct {
+	User  string `json:"user"`
+	Roles []Role `json:"roles"`
+}
+
 func v2AuthURL(ep url.URL, action string, name string) *url.URL {
 	if name != "" {
 		ep.Path = path.Join(ep.Path, defaultV2AuthPrefix, action, name)
@@ -78,9 +89,9 @@ func (s *httpAuthAPI) enableDisable(ctx context.Context, req httpAction) error {
 	if err != nil {
 		return err
 	}
-	if err := assertStatusCode(resp.StatusCode, http.StatusOK, http.StatusCreated); err != nil {
+	if err = assertStatusCode(resp.StatusCode, http.StatusOK, http.StatusCreated); err != nil {
 		var sec authError
-		err := json.Unmarshal(body, &sec)
+		err = json.Unmarshal(body, &sec)
 		if err != nil {
 			return err
 		}
@@ -117,25 +128,25 @@ func NewAuthUserAPI(c Client) AuthUserAPI {
 }
 
 type AuthUserAPI interface {
-	// Add a user.
+	// AddUser adds a user.
 	AddUser(ctx context.Context, username string, password string) error
 
-	// Remove a user.
+	// RemoveUser removes a user.
 	RemoveUser(ctx context.Context, username string) error
 
-	// Get user details.
+	// GetUser retrieves user details.
 	GetUser(ctx context.Context, username string) (*User, error)
 
-	// Grant a user some permission roles.
+	// GrantUser grants a user some permission roles.
 	GrantUser(ctx context.Context, username string, roles []string) (*User, error)
 
-	// Revoke some permission roles from a user.
+	// RevokeUser revokes some permission roles from a user.
 	RevokeUser(ctx context.Context, username string, roles []string) (*User, error)
 
-	// Change the user's password.
+	// ChangePassword changes the user's password.
 	ChangePassword(ctx context.Context, username string, password string) (*User, error)
 
-	// List users.
+	// ListUsers lists the users.
 	ListUsers(ctx context.Context) ([]string, error)
 }
 
@@ -179,22 +190,28 @@ func (u *httpAuthUserAPI) ListUsers(ctx context.Context) ([]string, error) {
 	if err != nil {
 		return nil, err
 	}
-	if err := assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
+	if err = assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
 		var sec authError
-		err := json.Unmarshal(body, &sec)
+		err = json.Unmarshal(body, &sec)
 		if err != nil {
 			return nil, err
 		}
 		return nil, sec
 	}
+
 	var userList struct {
-		Users []string `json:"users"`
+		Users []userListEntry `json:"users"`
 	}
-	err = json.Unmarshal(body, &userList)
+
-	if err != nil {
+	if err = json.Unmarshal(body, &userList); err != nil {
 		return nil, err
 	}
-	return userList.Users, nil
+
+	ret := make([]string, 0, len(userList.Users))
+	for _, u := range userList.Users {
+		ret = append(ret, u.User)
+	}
+	return ret, nil
 }
 
 func (u *httpAuthUserAPI) AddUser(ctx context.Context, username string, password string) error {
@@ -221,9 +238,9 @@ func (u *httpAuthUserAPI) addRemoveUser(ctx context.Context, req *authUserAPIAct
 	if err != nil {
 		return err
 	}
-	if err := assertStatusCode(resp.StatusCode, http.StatusOK, http.StatusCreated); err != nil {
+	if err = assertStatusCode(resp.StatusCode, http.StatusOK, http.StatusCreated); err != nil {
 		var sec authError
-		err := json.Unmarshal(body, &sec)
+		err = json.Unmarshal(body, &sec)
 		if err != nil {
 			return err
 		}
@@ -280,18 +297,24 @@ func (u *httpAuthUserAPI) modUser(ctx context.Context, req *authUserAPIAction) (
 	if err != nil {
 		return nil, err
 	}
-	if err := assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
+	if err = assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
 		var sec authError
-		err := json.Unmarshal(body, &sec)
+		err = json.Unmarshal(body, &sec)
 		if err != nil {
 			return nil, err
 		}
 		return nil, sec
 	}
 	var user User
-	err = json.Unmarshal(body, &user)
+	if err = json.Unmarshal(body, &user); err != nil {
-	if err != nil {
+		var userR UserRoles
-		return nil, err
+		if urerr := json.Unmarshal(body, &userR); urerr != nil {
+			return nil, err
+		}
+		user.User = userR.User
+		for _, r := range userR.Roles {
+			user.Roles = append(user.Roles, r.Role)
+		}
 	}
 	return &user, nil
 }

vendor/github.com/coreos/etcd/client/cancelreq.go

@@ -4,8 +4,6 @@
 
 // borrowed from golang/net/context/ctxhttp/cancelreq.go
 
-// +build go1.5
-
 package client
 
 import "net/http"

vendor/github.com/coreos/etcd/client/cancelreq_go14.go

@@ -1,17 +0,0 @@
-// Copyright 2015 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// borrowed from golang/net/context/ctxhttp/cancelreq_go14.go
-
-// +build !go1.5
-
-package client
-
-import "net/http"
-
-func requestCanceler(tr CancelableTransport, req *http.Request) func() {
-	return func() {
-		tr.CancelRequest(req)
-	}
-}

vendor/github.com/coreos/etcd/client/client.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -15,6 +15,7 @@
 package client
 
 import (
+	"encoding/json"
 	"errors"
 	"fmt"
 	"io/ioutil"
@@ -22,11 +23,13 @@ import (
 	"net"
 	"net/http"
 	"net/url"
-	"reflect"
 	"sort"
+	"strconv"
 	"sync"
 	"time"
 
+	"github.com/coreos/etcd/version"
+
 	"golang.org/x/net/context"
 )
 
@@ -34,7 +37,12 @@ var (
 	ErrNoEndpoints           = errors.New("client: no endpoints available")
 	ErrTooManyRedirects      = errors.New("client: too many redirects")
 	ErrClusterUnavailable    = errors.New("client: etcd cluster is unavailable or misconfigured")
+	ErrNoLeaderEndpoint      = errors.New("client: no leader endpoint available")
 	errTooManyRedirectChecks = errors.New("client: too many redirect checks")
+
+	// oneShotCtxValue is set on a context using WithValue(&oneShotValue) so
+	// that Do() will not retry a request
+	oneShotCtxValue interface{}
 )
 
 var DefaultRequestTimeout = 5 * time.Second
@@ -48,6 +56,29 @@ var DefaultTransport CancelableTransport = &http.Transport{
 	TLSHandshakeTimeout: 10 * time.Second,
 }
 
+type EndpointSelectionMode int
+
+const (
+	// EndpointSelectionRandom is the default value of the 'SelectionMode'.
+	// As the name implies, the client object will pick a node from the members
+	// of the cluster in a random fashion. If the cluster has three members, A, B,
+	// and C, the client picks any node from its three members as its request
+	// destination.
+	EndpointSelectionRandom EndpointSelectionMode = iota
+
+	// If 'SelectionMode' is set to 'EndpointSelectionPrioritizeLeader',
+	// requests are sent directly to the cluster leader. This reduces
+	// forwarding roundtrips compared to making requests to etcd followers
+	// who then forward them to the cluster leader. In the event of a leader
+	// failure, however, clients configured this way cannot prioritize among
+	// the remaining etcd followers. Therefore, when a client sets 'SelectionMode'
+	// to 'EndpointSelectionPrioritizeLeader', it must use 'client.AutoSync()' to
+	// maintain its knowledge of current cluster state.
+	//
+	// This mode should be used with Client.AutoSync().
+	EndpointSelectionPrioritizeLeader
+)
+
 type Config struct {
 	// Endpoints defines a set of URLs (schemes, hosts and ports only)
 	// that can be used to communicate with a logical etcd cluster. For
@@ -73,7 +104,7 @@ type Config struct {
 	// CheckRedirect specifies the policy for handling HTTP redirects.
 	// If CheckRedirect is not nil, the Client calls it before
 	// following an HTTP redirect. The sole argument is the number of
-	// requests that have alrady been made. If CheckRedirect returns
+	// requests that have already been made. If CheckRedirect returns
 	// an error, Client.Do will not make any further requests and return
 	// the error back it to the caller.
 	//
@@ -99,11 +130,17 @@ type Config struct {
 	// watch start. But if server is behind some kind of proxy, the response
 	// header may be cached at proxy, and Client cannot rely on this behavior.
 	//
+	// Especially, wait request will ignore this timeout.
+	//
 	// One API call may send multiple requests to different etcd servers until it
 	// succeeds. Use context of the API to specify the overall timeout.
 	//
 	// A HeaderTimeoutPerRequest of zero means no timeout.
 	HeaderTimeoutPerRequest time.Duration
+
+	// SelectionMode is an EndpointSelectionMode enum that specifies the
+	// policy for choosing the etcd cluster node to which requests are sent.
+	SelectionMode EndpointSelectionMode
 }
 
 func (cfg *Config) transport() CancelableTransport {
@@ -162,6 +199,14 @@ type Client interface {
 	// this may differ from the initial Endpoints provided in the Config.
 	Endpoints() []string
 
+	// SetEndpoints sets the set of API endpoints used by Client to resolve
+	// HTTP requests. If the given endpoints are not valid, an error will be
+	// returned
+	SetEndpoints(eps []string) error
+
+	// GetVersion retrieves the current etcd server and cluster version
+	GetVersion(ctx context.Context) (*version.Versions, error)
+
 	httpClient
 }
 
@@ -169,6 +214,7 @@ func New(cfg Config) (Client, error) {
 	c := &httpClusterClient{
 		clientFactory: newHTTPClientFactory(cfg.transport(), cfg.checkRedirect(), cfg.HeaderTimeoutPerRequest),
 		rand:          rand.New(rand.NewSource(int64(time.Now().Nanosecond()))),
+		selectionMode: cfg.SelectionMode,
 	}
 	if cfg.Username != "" {
 		c.credentials = &credentials{
@@ -176,7 +222,7 @@ func New(cfg Config) (Client, error) {
 			password: cfg.Password,
 		}
 	}
-	if err := c.reset(cfg.Endpoints); err != nil {
+	if err := c.SetEndpoints(cfg.Endpoints); err != nil {
 		return nil, err
 	}
 	return c, nil
@@ -216,25 +262,69 @@ type httpClusterClient struct {
 	pinned        int
 	credentials   *credentials
 	sync.RWMutex
-	rand *rand.Rand
+	rand          *rand.Rand
+	selectionMode EndpointSelectionMode
 }
 
-func (c *httpClusterClient) reset(eps []string) error {
+func (c *httpClusterClient) getLeaderEndpoint(ctx context.Context, eps []url.URL) (string, error) {
+	ceps := make([]url.URL, len(eps))
+	copy(ceps, eps)
+
+	// To perform a lookup on the new endpoint list without using the current
+	// client, we'll copy it
+	clientCopy := &httpClusterClient{
+		clientFactory: c.clientFactory,
+		credentials:   c.credentials,
+		rand:          c.rand,
+
+		pinned:    0,
+		endpoints: ceps,
+	}
+
+	mAPI := NewMembersAPI(clientCopy)
+	leader, err := mAPI.Leader(ctx)
+	if err != nil {
+		return "", err
+	}
+	if len(leader.ClientURLs) == 0 {
+		return "", ErrNoLeaderEndpoint
+	}
+
+	return leader.ClientURLs[0], nil // TODO: how to handle multiple client URLs?
+}
+
+func (c *httpClusterClient) parseEndpoints(eps []string) ([]url.URL, error) {
 	if len(eps) == 0 {
-		return ErrNoEndpoints
+		return []url.URL{}, ErrNoEndpoints
 	}
 
 	neps := make([]url.URL, len(eps))
 	for i, ep := range eps {
 		u, err := url.Parse(ep)
 		if err != nil {
-			return err
+			return []url.URL{}, err
 		}
 		neps[i] = *u
 	}
+	return neps, nil
+}
+
+func (c *httpClusterClient) SetEndpoints(eps []string) error {
+	neps, err := c.parseEndpoints(eps)
+	if err != nil {
+		return err
+	}
+
+	c.Lock()
+	defer c.Unlock()
 
 	c.endpoints = shuffleEndpoints(c.rand, neps)
-	// TODO: pin old endpoint if possible, and rebalance when new endpoint appears
+	// We're not doing anything for PrioritizeLeader here. This is
+	// due to not having a context meaning we can't call getLeaderEndpoint
+	// However, if you're using PrioritizeLeader, you've already been told
+	// to regularly call sync, where we do have a ctx, and can figure the
+	// leader. PrioritizeLeader is also quite a loose guarantee, so deal
+	// with it
 	c.pinned = 0
 
 	return nil
@@ -268,6 +358,7 @@ func (c *httpClusterClient) Do(ctx context.Context, act httpAction) (*http.Respo
 	var body []byte
 	var err error
 	cerr := &ClusterError{}
+	isOneShot := ctx.Value(&oneShotCtxValue) != nil
 
 	for i := pinned; i < leps+pinned; i++ {
 		k := i % leps
@@ -275,13 +366,13 @@ func (c *httpClusterClient) Do(ctx context.Context, act httpAction) (*http.Respo
 		resp, body, err = hc.Do(ctx, action)
 		if err != nil {
 			cerr.Errors = append(cerr.Errors, err)
-			// mask previous errors with context error, which is controlled by user
+			if err == ctx.Err() {
+				return nil, nil, ctx.Err()
+			}
 			if err == context.Canceled || err == context.DeadlineExceeded {
 				return nil, nil, err
 			}
-			continue
+		} else if resp.StatusCode/100 == 5 {
-		}
-		if resp.StatusCode/100 == 5 {
 			switch resp.StatusCode {
 			case http.StatusInternalServerError, http.StatusServiceUnavailable:
 				// TODO: make sure this is a no leader response
@@ -289,7 +380,16 @@ func (c *httpClusterClient) Do(ctx context.Context, act httpAction) (*http.Respo
 			default:
 				cerr.Errors = append(cerr.Errors, fmt.Errorf("client: etcd member %s returns server error [%s]", eps[k].String(), http.StatusText(resp.StatusCode)))
 			}
-			continue
+			err = cerr.Errors[0]
+		}
+		if err != nil {
+			if !isOneShot {
+				continue
+			}
+			c.Lock()
+			c.pinned = (k + 1) % leps
+			c.Unlock()
+			return nil, nil, err
 		}
 		if k != pinned {
 			c.Lock()
@@ -321,27 +421,51 @@ func (c *httpClusterClient) Sync(ctx context.Context) error {
 		return err
 	}
 
-	c.Lock()
+	var eps []string
-	defer c.Unlock()
-
-	eps := make([]string, 0)
 	for _, m := range ms {
 		eps = append(eps, m.ClientURLs...)
 	}
-	sort.Sort(sort.StringSlice(eps))
 
-	ceps := make([]string, len(c.endpoints))
+	neps, err := c.parseEndpoints(eps)
-	for i, cep := range c.endpoints {
+	if err != nil {
-		ceps[i] = cep.String()
+		return err
-	}
-	sort.Sort(sort.StringSlice(ceps))
-	// fast path if no change happens
-	// this helps client to pin the endpoint when no cluster change
-	if reflect.DeepEqual(eps, ceps) {
-		return nil
 	}
 
-	return c.reset(eps)
+	npin := 0
+
+	switch c.selectionMode {
+	case EndpointSelectionRandom:
+		c.RLock()
+		eq := endpointsEqual(c.endpoints, neps)
+		c.RUnlock()
+
+		if eq {
+			return nil
+		}
+		// When items in the endpoint list changes, we choose a new pin
+		neps = shuffleEndpoints(c.rand, neps)
+	case EndpointSelectionPrioritizeLeader:
+		nle, err := c.getLeaderEndpoint(ctx, neps)
+		if err != nil {
+			return ErrNoLeaderEndpoint
+		}
+
+		for i, n := range neps {
+			if n.String() == nle {
+				npin = i
+				break
+			}
+		}
+	default:
+		return fmt.Errorf("invalid endpoint selection mode: %d", c.selectionMode)
+	}
+
+	c.Lock()
+	defer c.Unlock()
+	c.endpoints = neps
+	c.pinned = npin
+
+	return nil
 }
 
 func (c *httpClusterClient) AutoSync(ctx context.Context, interval time.Duration) error {
@@ -360,6 +484,33 @@ func (c *httpClusterClient) AutoSync(ctx context.Context, interval time.Duration
 	}
 }
 
+func (c *httpClusterClient) GetVersion(ctx context.Context) (*version.Versions, error) {
+	act := &getAction{Prefix: "/version"}
+
+	resp, body, err := c.Do(ctx, act)
+	if err != nil {
+		return nil, err
+	}
+
+	switch resp.StatusCode {
+	case http.StatusOK:
+		if len(body) == 0 {
+			return nil, ErrEmptyBody
+		}
+		var vresp version.Versions
+		if err := json.Unmarshal(body, &vresp); err != nil {
+			return nil, ErrInvalidJSON
+		}
+		return &vresp, nil
+	default:
+		var etcdErr Error
+		if err := json.Unmarshal(body, &etcdErr); err != nil {
+			return nil, ErrInvalidJSON
+		}
+		return nil, etcdErr
+	}
+}
+
 type roundTripResponse struct {
 	resp *http.Response
 	err  error
@@ -378,9 +529,24 @@ func (c *simpleHTTPClient) Do(ctx context.Context, act httpAction) (*http.Respon
 		return nil, nil, err
 	}
 
-	hctx, hcancel := context.WithCancel(ctx)
+	isWait := false
-	if c.headerTimeout > 0 {
+	if req != nil && req.URL != nil {
+		ws := req.URL.Query().Get("wait")
+		if len(ws) != 0 {
+			var err error
+			isWait, err = strconv.ParseBool(ws)
+			if err != nil {
+				return nil, nil, fmt.Errorf("wrong wait value %s (%v for %+v)", ws, err, req)
+			}
+		}
+	}
+
+	var hctx context.Context
+	var hcancel context.CancelFunc
+	if !isWait && c.headerTimeout > 0 {
 		hctx, hcancel = context.WithTimeout(ctx, c.headerTimeout)
+	} else {
+		hctx, hcancel = context.WithCancel(ctx)
 	}
 	defer hcancel()
 
@@ -512,3 +678,27 @@ func shuffleEndpoints(r *rand.Rand, eps []url.URL) []url.URL {
 	}
 	return neps
 }
+
+func endpointsEqual(left, right []url.URL) bool {
+	if len(left) != len(right) {
+		return false
+	}
+
+	sLeft := make([]string, len(left))
+	sRight := make([]string, len(right))
+	for i, l := range left {
+		sLeft[i] = l.String()
+	}
+	for i, r := range right {
+		sRight[i] = r.String()
+	}
+
+	sort.Strings(sLeft)
+	sort.Strings(sRight)
+	for i := range sLeft {
+		if sLeft[i] != sRight[i] {
+			return false
+		}
+	}
+	return true
+}

vendor/github.com/coreos/etcd/client/cluster_error.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -21,7 +21,11 @@ type ClusterError struct {
 }
 
 func (ce *ClusterError) Error() string {
-	return ErrClusterUnavailable.Error()
+	s := ErrClusterUnavailable.Error()
+	for i, e := range ce.Errors {
+		s += fmt.Sprintf("; error #%d: %s\n", i, e)
+	}
+	return s
 }
 
 func (ce *ClusterError) Detail() string {

vendor/github.com/coreos/etcd/client/curl.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.

vendor/github.com/coreos/etcd/client/discover.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -14,8 +14,27 @@
 
 package client
 
+import (
+	"github.com/coreos/etcd/pkg/srv"
+)
+
 // Discoverer is an interface that wraps the Discover method.
 type Discoverer interface {
 	// Discover looks up the etcd servers for the domain.
 	Discover(domain string) ([]string, error)
 }
+
+type srvDiscover struct{}
+
+// NewSRVDiscover constructs a new Discoverer that uses the stdlib to lookup SRV records.
+func NewSRVDiscover() Discoverer {
+	return &srvDiscover{}
+}
+
+func (d *srvDiscover) Discover(domain string) ([]string, error) {
+	srvs, err := srv.GetClient("etcd-client", domain)
+	if err != nil {
+		return nil, err
+	}
+	return srvs.Endpoints, nil
+}

vendor/github.com/coreos/etcd/client/doc.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -34,6 +34,8 @@ Create a Config and exchange it for a Client:
 		// handle error
 	}
 
+Clients are safe for concurrent use by multiple goroutines.
+
 Create a KeysAPI using the Client, then use it to interact with etcd:
 
 	kAPI := client.NewKeysAPI(c)

vendor/github.com/coreos/etcd/client/keys.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -106,7 +106,7 @@ type KeysAPI interface {
 
 	// Set assigns a new value to a Node identified by a given key. The caller
 	// may define a set of conditions in the SetOptions. If SetOptions.Dir=true
-	// than value is ignored.
+	// then value is ignored.
 	Set(ctx context.Context, key, value string, opts *SetOptions) (*Response, error)
 
 	// Delete removes a Node identified by the given key, optionally destroying
@@ -184,8 +184,17 @@ type SetOptions struct {
 	// a TTL of 0.
 	TTL time.Duration
 
+	// Refresh set to true means a TTL value can be updated
+	// without firing a watch or changing the node value. A
+	// value must not be provided when refreshing a key.
+	Refresh bool
+
 	// Dir specifies whether or not this Node should be created as a directory.
 	Dir bool
+
+	// NoValueOnSuccess specifies whether the response contains the current value of the Node.
+	// If set, the response will only contain the current value when the request fails.
+	NoValueOnSuccess bool
 }
 
 type GetOptions struct {
@@ -234,7 +243,7 @@ type DeleteOptions struct {
 
 type Watcher interface {
 	// Next blocks until an etcd event occurs, then returns a Response
-	// represeting that event. The behavior of Next depends on the
+	// representing that event. The behavior of Next depends on the
 	// WatcherOptions used to construct the Watcher. Next is designed to
 	// be called repeatedly, each time blocking until a subsequent event
 	// is available.
@@ -263,6 +272,10 @@ type Response struct {
 	// Index holds the cluster-level index at the time the Response was generated.
 	// This index is not tied to the Node(s) contained in this Response.
 	Index uint64 `json:"-"`
+
+	// ClusterID holds the cluster-level ID reported by the server.  This
+	// should be different for different etcd clusters.
+	ClusterID string `json:"-"`
 }
 
 type Node struct {
@@ -306,6 +319,7 @@ func (n *Node) TTLDuration() time.Duration {
 type Nodes []*Node
 
 // interfaces for sorting
+
 func (ns Nodes) Len() int           { return len(ns) }
 func (ns Nodes) Less(i, j int) bool { return ns[i].Key < ns[j].Key }
 func (ns Nodes) Swap(i, j int)      { ns[i], ns[j] = ns[j], ns[i] }
@@ -327,10 +341,16 @@ func (k *httpKeysAPI) Set(ctx context.Context, key, val string, opts *SetOptions
 		act.PrevIndex = opts.PrevIndex
 		act.PrevExist = opts.PrevExist
 		act.TTL = opts.TTL
+		act.Refresh = opts.Refresh
 		act.Dir = opts.Dir
+		act.NoValueOnSuccess = opts.NoValueOnSuccess
 	}
 
-	resp, body, err := k.client.Do(ctx, act)
+	doCtx := ctx
+	if act.PrevExist == PrevNoExist {
+		doCtx = context.WithValue(doCtx, &oneShotCtxValue, &oneShotCtxValue)
+	}
+	resp, body, err := k.client.Do(doCtx, act)
 	if err != nil {
 		return nil, err
 	}
@@ -378,7 +398,8 @@ func (k *httpKeysAPI) Delete(ctx context.Context, key string, opts *DeleteOption
 		act.Recursive = opts.Recursive
 	}
 
-	resp, body, err := k.client.Do(ctx, act)
+	doCtx := context.WithValue(ctx, &oneShotCtxValue, &oneShotCtxValue)
+	resp, body, err := k.client.Do(doCtx, act)
 	if err != nil {
 		return nil, err
 	}
@@ -511,14 +532,16 @@ func (w *waitAction) HTTPRequest(ep url.URL) *http.Request {
 }
 
 type setAction struct {
-	Prefix    string
+	Prefix           string
-	Key       string
+	Key              string
-	Value     string
+	Value            string
-	PrevValue string
+	PrevValue        string
-	PrevIndex uint64
+	PrevIndex        uint64
-	PrevExist PrevExistType
+	PrevExist        PrevExistType
-	TTL       time.Duration
+	TTL              time.Duration
-	Dir       bool
+	Refresh          bool
+	Dir              bool
+	NoValueOnSuccess bool
 }
 
 func (a *setAction) HTTPRequest(ep url.URL) *http.Request {
@@ -549,6 +572,13 @@ func (a *setAction) HTTPRequest(ep url.URL) *http.Request {
 		form.Add("ttl", strconv.FormatUint(uint64(a.TTL.Seconds()), 10))
 	}
 
+	if a.Refresh {
+		form.Add("refresh", "true")
+	}
+	if a.NoValueOnSuccess {
+		params.Set("noValueOnSuccess", strconv.FormatBool(a.NoValueOnSuccess))
+	}
+
 	u.RawQuery = params.Encode()
 	body := strings.NewReader(form.Encode())
 
@@ -639,6 +669,7 @@ func unmarshalSuccessfulKeysResponse(header http.Header, body []byte) (*Response
 			return nil, err
 		}
 	}
+	res.ClusterID = header.Get("X-Etcd-Cluster-ID")
 	return &res, nil
 }
 

vendor/github.com/coreos/etcd/client/members.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -29,6 +29,7 @@ import (
 
 var (
 	defaultV2MembersPrefix = "/v2/members"
+	defaultLeaderSuffix    = "/leader"
 )
 
 type Member struct {
@@ -105,6 +106,9 @@ type MembersAPI interface {
 
 	// Update instructs etcd to update an existing Member in the cluster.
 	Update(ctx context.Context, mID string, peerURLs []string) error
+
+	// Leader gets current leader of the cluster
+	Leader(ctx context.Context) (*Member, error)
 }
 
 type httpMembersAPI struct {
@@ -199,6 +203,25 @@ func (m *httpMembersAPI) Remove(ctx context.Context, memberID string) error {
 	return assertStatusCode(resp.StatusCode, http.StatusNoContent, http.StatusGone)
 }
 
+func (m *httpMembersAPI) Leader(ctx context.Context) (*Member, error) {
+	req := &membersAPIActionLeader{}
+	resp, body, err := m.client.Do(ctx, req)
+	if err != nil {
+		return nil, err
+	}
+
+	if err := assertStatusCode(resp.StatusCode, http.StatusOK); err != nil {
+		return nil, err
+	}
+
+	var leader Member
+	if err := json.Unmarshal(body, &leader); err != nil {
+		return nil, err
+	}
+
+	return &leader, nil
+}
+
 type membersAPIActionList struct{}
 
 func (l *membersAPIActionList) HTTPRequest(ep url.URL) *http.Request {
@@ -255,6 +278,15 @@ func assertStatusCode(got int, want ...int) (err error) {
 	return fmt.Errorf("unexpected status code %d", got)
 }
 
+type membersAPIActionLeader struct{}
+
+func (l *membersAPIActionLeader) HTTPRequest(ep url.URL) *http.Request {
+	u := v2MembersURL(ep)
+	u.Path = path.Join(u.Path, defaultLeaderSuffix)
+	req, _ := http.NewRequest("GET", u.String(), nil)
+	return req
+}
+
 // v2MembersURL add the necessary path to the provided endpoint
 // to route requests to the default v2 members API.
 func v2MembersURL(ep url.URL) *url.URL {

vendor/github.com/coreos/etcd/client/srv.go

@@ -1,65 +0,0 @@
-// Copyright 2015 CoreOS, Inc.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-package client
-
-import (
-	"fmt"
-	"net"
-	"net/url"
-)
-
-var (
-	// indirection for testing
-	lookupSRV = net.LookupSRV
-)
-
-type srvDiscover struct{}
-
-// NewSRVDiscover constructs a new Dicoverer that uses the stdlib to lookup SRV records.
-func NewSRVDiscover() Discoverer {
-	return &srvDiscover{}
-}
-
-// Discover looks up the etcd servers for the domain.
-func (d *srvDiscover) Discover(domain string) ([]string, error) {
-	var urls []*url.URL
-
-	updateURLs := func(service, scheme string) error {
-		_, addrs, err := lookupSRV(service, "tcp", domain)
-		if err != nil {
-			return err
-		}
-		for _, srv := range addrs {
-			urls = append(urls, &url.URL{
-				Scheme: scheme,
-				Host:   net.JoinHostPort(srv.Target, fmt.Sprintf("%d", srv.Port)),
-			})
-		}
-		return nil
-	}
-
-	errHTTPS := updateURLs("etcd-server-ssl", "https")
-	errHTTP := updateURLs("etcd-server", "http")
-
-	if errHTTPS != nil && errHTTP != nil {
-		return nil, fmt.Errorf("dns lookup errors: %s and %s", errHTTPS, errHTTP)
-	}
-
-	endpoints := make([]string, len(urls))
-	for i := range urls {
-		endpoints[i] = urls[i].String()
-	}
-	return endpoints, nil
-}

vendor/github.com/coreos/etcd/client/util.go

@@ -0,0 +1,53 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package client
+
+import (
+	"regexp"
+)
+
+var (
+	roleNotFoundRegExp *regexp.Regexp
+	userNotFoundRegExp *regexp.Regexp
+)
+
+func init() {
+	roleNotFoundRegExp = regexp.MustCompile("auth: Role .* does not exist.")
+	userNotFoundRegExp = regexp.MustCompile("auth: User .* does not exist.")
+}
+
+// IsKeyNotFound returns true if the error code is ErrorCodeKeyNotFound.
+func IsKeyNotFound(err error) bool {
+	if cErr, ok := err.(Error); ok {
+		return cErr.Code == ErrorCodeKeyNotFound
+	}
+	return false
+}
+
+// IsRoleNotFound returns true if the error means role not found of v2 API.
+func IsRoleNotFound(err error) bool {
+	if ae, ok := err.(authError); ok {
+		return roleNotFoundRegExp.MatchString(ae.Message)
+	}
+	return false
+}
+
+// IsUserNotFound returns true if the error means user not found of v2 API.
+func IsUserNotFound(err error) bool {
+	if ae, ok := err.(authError); ok {
+		return userNotFoundRegExp.MatchString(ae.Message)
+	}
+	return false
+}

vendor/github.com/coreos/etcd/clientv3/auth.go

@@ -0,0 +1,222 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/coreos/etcd/auth/authpb"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+)
+
+type (
+	AuthEnableResponse               pb.AuthEnableResponse
+	AuthDisableResponse              pb.AuthDisableResponse
+	AuthenticateResponse             pb.AuthenticateResponse
+	AuthUserAddResponse              pb.AuthUserAddResponse
+	AuthUserDeleteResponse           pb.AuthUserDeleteResponse
+	AuthUserChangePasswordResponse   pb.AuthUserChangePasswordResponse
+	AuthUserGrantRoleResponse        pb.AuthUserGrantRoleResponse
+	AuthUserGetResponse              pb.AuthUserGetResponse
+	AuthUserRevokeRoleResponse       pb.AuthUserRevokeRoleResponse
+	AuthRoleAddResponse              pb.AuthRoleAddResponse
+	AuthRoleGrantPermissionResponse  pb.AuthRoleGrantPermissionResponse
+	AuthRoleGetResponse              pb.AuthRoleGetResponse
+	AuthRoleRevokePermissionResponse pb.AuthRoleRevokePermissionResponse
+	AuthRoleDeleteResponse           pb.AuthRoleDeleteResponse
+	AuthUserListResponse             pb.AuthUserListResponse
+	AuthRoleListResponse             pb.AuthRoleListResponse
+
+	PermissionType authpb.Permission_Type
+	Permission     authpb.Permission
+)
+
+const (
+	PermRead      = authpb.READ
+	PermWrite     = authpb.WRITE
+	PermReadWrite = authpb.READWRITE
+)
+
+type Auth interface {
+	// AuthEnable enables auth of an etcd cluster.
+	AuthEnable(ctx context.Context) (*AuthEnableResponse, error)
+
+	// AuthDisable disables auth of an etcd cluster.
+	AuthDisable(ctx context.Context) (*AuthDisableResponse, error)
+
+	// UserAdd adds a new user to an etcd cluster.
+	UserAdd(ctx context.Context, name string, password string) (*AuthUserAddResponse, error)
+
+	// UserDelete deletes a user from an etcd cluster.
+	UserDelete(ctx context.Context, name string) (*AuthUserDeleteResponse, error)
+
+	// UserChangePassword changes a password of a user.
+	UserChangePassword(ctx context.Context, name string, password string) (*AuthUserChangePasswordResponse, error)
+
+	// UserGrantRole grants a role to a user.
+	UserGrantRole(ctx context.Context, user string, role string) (*AuthUserGrantRoleResponse, error)
+
+	// UserGet gets a detailed information of a user.
+	UserGet(ctx context.Context, name string) (*AuthUserGetResponse, error)
+
+	// UserList gets a list of all users.
+	UserList(ctx context.Context) (*AuthUserListResponse, error)
+
+	// UserRevokeRole revokes a role of a user.
+	UserRevokeRole(ctx context.Context, name string, role string) (*AuthUserRevokeRoleResponse, error)
+
+	// RoleAdd adds a new role to an etcd cluster.
+	RoleAdd(ctx context.Context, name string) (*AuthRoleAddResponse, error)
+
+	// RoleGrantPermission grants a permission to a role.
+	RoleGrantPermission(ctx context.Context, name string, key, rangeEnd string, permType PermissionType) (*AuthRoleGrantPermissionResponse, error)
+
+	// RoleGet gets a detailed information of a role.
+	RoleGet(ctx context.Context, role string) (*AuthRoleGetResponse, error)
+
+	// RoleList gets a list of all roles.
+	RoleList(ctx context.Context) (*AuthRoleListResponse, error)
+
+	// RoleRevokePermission revokes a permission from a role.
+	RoleRevokePermission(ctx context.Context, role string, key, rangeEnd string) (*AuthRoleRevokePermissionResponse, error)
+
+	// RoleDelete deletes a role.
+	RoleDelete(ctx context.Context, role string) (*AuthRoleDeleteResponse, error)
+}
+
+type auth struct {
+	remote pb.AuthClient
+}
+
+func NewAuth(c *Client) Auth {
+	return &auth{remote: pb.NewAuthClient(c.ActiveConnection())}
+}
+
+func (auth *auth) AuthEnable(ctx context.Context) (*AuthEnableResponse, error) {
+	resp, err := auth.remote.AuthEnable(ctx, &pb.AuthEnableRequest{}, grpc.FailFast(false))
+	return (*AuthEnableResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) AuthDisable(ctx context.Context) (*AuthDisableResponse, error) {
+	resp, err := auth.remote.AuthDisable(ctx, &pb.AuthDisableRequest{}, grpc.FailFast(false))
+	return (*AuthDisableResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserAdd(ctx context.Context, name string, password string) (*AuthUserAddResponse, error) {
+	resp, err := auth.remote.UserAdd(ctx, &pb.AuthUserAddRequest{Name: name, Password: password})
+	return (*AuthUserAddResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserDelete(ctx context.Context, name string) (*AuthUserDeleteResponse, error) {
+	resp, err := auth.remote.UserDelete(ctx, &pb.AuthUserDeleteRequest{Name: name})
+	return (*AuthUserDeleteResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserChangePassword(ctx context.Context, name string, password string) (*AuthUserChangePasswordResponse, error) {
+	resp, err := auth.remote.UserChangePassword(ctx, &pb.AuthUserChangePasswordRequest{Name: name, Password: password})
+	return (*AuthUserChangePasswordResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserGrantRole(ctx context.Context, user string, role string) (*AuthUserGrantRoleResponse, error) {
+	resp, err := auth.remote.UserGrantRole(ctx, &pb.AuthUserGrantRoleRequest{User: user, Role: role})
+	return (*AuthUserGrantRoleResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserGet(ctx context.Context, name string) (*AuthUserGetResponse, error) {
+	resp, err := auth.remote.UserGet(ctx, &pb.AuthUserGetRequest{Name: name}, grpc.FailFast(false))
+	return (*AuthUserGetResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserList(ctx context.Context) (*AuthUserListResponse, error) {
+	resp, err := auth.remote.UserList(ctx, &pb.AuthUserListRequest{}, grpc.FailFast(false))
+	return (*AuthUserListResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) UserRevokeRole(ctx context.Context, name string, role string) (*AuthUserRevokeRoleResponse, error) {
+	resp, err := auth.remote.UserRevokeRole(ctx, &pb.AuthUserRevokeRoleRequest{Name: name, Role: role})
+	return (*AuthUserRevokeRoleResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) RoleAdd(ctx context.Context, name string) (*AuthRoleAddResponse, error) {
+	resp, err := auth.remote.RoleAdd(ctx, &pb.AuthRoleAddRequest{Name: name})
+	return (*AuthRoleAddResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) RoleGrantPermission(ctx context.Context, name string, key, rangeEnd string, permType PermissionType) (*AuthRoleGrantPermissionResponse, error) {
+	perm := &authpb.Permission{
+		Key:      []byte(key),
+		RangeEnd: []byte(rangeEnd),
+		PermType: authpb.Permission_Type(permType),
+	}
+	resp, err := auth.remote.RoleGrantPermission(ctx, &pb.AuthRoleGrantPermissionRequest{Name: name, Perm: perm})
+	return (*AuthRoleGrantPermissionResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) RoleGet(ctx context.Context, role string) (*AuthRoleGetResponse, error) {
+	resp, err := auth.remote.RoleGet(ctx, &pb.AuthRoleGetRequest{Role: role}, grpc.FailFast(false))
+	return (*AuthRoleGetResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) RoleList(ctx context.Context) (*AuthRoleListResponse, error) {
+	resp, err := auth.remote.RoleList(ctx, &pb.AuthRoleListRequest{}, grpc.FailFast(false))
+	return (*AuthRoleListResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) RoleRevokePermission(ctx context.Context, role string, key, rangeEnd string) (*AuthRoleRevokePermissionResponse, error) {
+	resp, err := auth.remote.RoleRevokePermission(ctx, &pb.AuthRoleRevokePermissionRequest{Role: role, Key: key, RangeEnd: rangeEnd})
+	return (*AuthRoleRevokePermissionResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *auth) RoleDelete(ctx context.Context, role string) (*AuthRoleDeleteResponse, error) {
+	resp, err := auth.remote.RoleDelete(ctx, &pb.AuthRoleDeleteRequest{Role: role})
+	return (*AuthRoleDeleteResponse)(resp), toErr(ctx, err)
+}
+
+func StrToPermissionType(s string) (PermissionType, error) {
+	val, ok := authpb.Permission_Type_value[strings.ToUpper(s)]
+	if ok {
+		return PermissionType(val), nil
+	}
+	return PermissionType(-1), fmt.Errorf("invalid permission type: %s", s)
+}
+
+type authenticator struct {
+	conn   *grpc.ClientConn // conn in-use
+	remote pb.AuthClient
+}
+
+func (auth *authenticator) authenticate(ctx context.Context, name string, password string) (*AuthenticateResponse, error) {
+	resp, err := auth.remote.Authenticate(ctx, &pb.AuthenticateRequest{Name: name, Password: password}, grpc.FailFast(false))
+	return (*AuthenticateResponse)(resp), toErr(ctx, err)
+}
+
+func (auth *authenticator) close() {
+	auth.conn.Close()
+}
+
+func newAuthenticator(endpoint string, opts []grpc.DialOption) (*authenticator, error) {
+	conn, err := grpc.Dial(endpoint, opts...)
+	if err != nil {
+		return nil, err
+	}
+
+	return &authenticator{
+		conn:   conn,
+		remote: pb.NewAuthClient(conn),
+	}, nil
+}

vendor/github.com/coreos/etcd/clientv3/balancer.go

@@ -0,0 +1,356 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"net/url"
+	"strings"
+	"sync"
+
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+)
+
+// ErrNoAddrAvilable is returned by Get() when the balancer does not have
+// any active connection to endpoints at the time.
+// This error is returned only when opts.BlockingWait is true.
+var ErrNoAddrAvilable = grpc.Errorf(codes.Unavailable, "there is no address available")
+
+// simpleBalancer does the bare minimum to expose multiple eps
+// to the grpc reconnection code path
+type simpleBalancer struct {
+	// addrs are the client's endpoints for grpc
+	addrs []grpc.Address
+	// notifyCh notifies grpc of the set of addresses for connecting
+	notifyCh chan []grpc.Address
+
+	// readyc closes once the first connection is up
+	readyc    chan struct{}
+	readyOnce sync.Once
+
+	// mu protects upEps, pinAddr, and connectingAddr
+	mu sync.RWMutex
+
+	// upc closes when upEps transitions from empty to non-zero or the balancer closes.
+	upc chan struct{}
+
+	// downc closes when grpc calls down() on pinAddr
+	downc chan struct{}
+
+	// stopc is closed to signal updateNotifyLoop should stop.
+	stopc chan struct{}
+
+	// donec closes when all goroutines are exited
+	donec chan struct{}
+
+	// updateAddrsC notifies updateNotifyLoop to update addrs.
+	updateAddrsC chan struct{}
+
+	// grpc issues TLS cert checks using the string passed into dial so
+	// that string must be the host. To recover the full scheme://host URL,
+	// have a map from hosts to the original endpoint.
+	host2ep map[string]string
+
+	// pinAddr is the currently pinned address; set to the empty string on
+	// intialization and shutdown.
+	pinAddr string
+
+	closed bool
+}
+
+func newSimpleBalancer(eps []string) *simpleBalancer {
+	notifyCh := make(chan []grpc.Address, 1)
+	addrs := make([]grpc.Address, len(eps))
+	for i := range eps {
+		addrs[i].Addr = getHost(eps[i])
+	}
+	sb := &simpleBalancer{
+		addrs:        addrs,
+		notifyCh:     notifyCh,
+		readyc:       make(chan struct{}),
+		upc:          make(chan struct{}),
+		stopc:        make(chan struct{}),
+		downc:        make(chan struct{}),
+		donec:        make(chan struct{}),
+		updateAddrsC: make(chan struct{}, 1),
+		host2ep:      getHost2ep(eps),
+	}
+	close(sb.downc)
+	go sb.updateNotifyLoop()
+	return sb
+}
+
+func (b *simpleBalancer) Start(target string, config grpc.BalancerConfig) error { return nil }
+
+func (b *simpleBalancer) ConnectNotify() <-chan struct{} {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return b.upc
+}
+
+func (b *simpleBalancer) getEndpoint(host string) string {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	return b.host2ep[host]
+}
+
+func getHost2ep(eps []string) map[string]string {
+	hm := make(map[string]string, len(eps))
+	for i := range eps {
+		_, host, _ := parseEndpoint(eps[i])
+		hm[host] = eps[i]
+	}
+	return hm
+}
+
+func (b *simpleBalancer) updateAddrs(eps []string) {
+	np := getHost2ep(eps)
+
+	b.mu.Lock()
+
+	match := len(np) == len(b.host2ep)
+	for k, v := range np {
+		if b.host2ep[k] != v {
+			match = false
+			break
+		}
+	}
+	if match {
+		// same endpoints, so no need to update address
+		b.mu.Unlock()
+		return
+	}
+
+	b.host2ep = np
+
+	addrs := make([]grpc.Address, 0, len(eps))
+	for i := range eps {
+		addrs = append(addrs, grpc.Address{Addr: getHost(eps[i])})
+	}
+	b.addrs = addrs
+
+	// updating notifyCh can trigger new connections,
+	// only update addrs if all connections are down
+	// or addrs does not include pinAddr.
+	update := !hasAddr(addrs, b.pinAddr)
+	b.mu.Unlock()
+
+	if update {
+		select {
+		case b.updateAddrsC <- struct{}{}:
+		case <-b.stopc:
+		}
+	}
+}
+
+func hasAddr(addrs []grpc.Address, targetAddr string) bool {
+	for _, addr := range addrs {
+		if targetAddr == addr.Addr {
+			return true
+		}
+	}
+	return false
+}
+
+func (b *simpleBalancer) updateNotifyLoop() {
+	defer close(b.donec)
+
+	for {
+		b.mu.RLock()
+		upc, downc, addr := b.upc, b.downc, b.pinAddr
+		b.mu.RUnlock()
+		// downc or upc should be closed
+		select {
+		case <-downc:
+			downc = nil
+		default:
+		}
+		select {
+		case <-upc:
+			upc = nil
+		default:
+		}
+		switch {
+		case downc == nil && upc == nil:
+			// stale
+			select {
+			case <-b.stopc:
+				return
+			default:
+			}
+		case downc == nil:
+			b.notifyAddrs()
+			select {
+			case <-upc:
+			case <-b.updateAddrsC:
+				b.notifyAddrs()
+			case <-b.stopc:
+				return
+			}
+		case upc == nil:
+			select {
+			// close connections that are not the pinned address
+			case b.notifyCh <- []grpc.Address{{Addr: addr}}:
+			case <-downc:
+			case <-b.stopc:
+				return
+			}
+			select {
+			case <-downc:
+			case <-b.updateAddrsC:
+			case <-b.stopc:
+				return
+			}
+			b.notifyAddrs()
+		}
+	}
+}
+
+func (b *simpleBalancer) notifyAddrs() {
+	b.mu.RLock()
+	addrs := b.addrs
+	b.mu.RUnlock()
+	select {
+	case b.notifyCh <- addrs:
+	case <-b.stopc:
+	}
+}
+
+func (b *simpleBalancer) Up(addr grpc.Address) func(error) {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+
+	// gRPC might call Up after it called Close. We add this check
+	// to "fix" it up at application layer. Or our simplerBalancer
+	// might panic since b.upc is closed.
+	if b.closed {
+		return func(err error) {}
+	}
+	// gRPC might call Up on a stale address.
+	// Prevent updating pinAddr with a stale address.
+	if !hasAddr(b.addrs, addr.Addr) {
+		return func(err error) {}
+	}
+	if b.pinAddr != "" {
+		return func(err error) {}
+	}
+	// notify waiting Get()s and pin first connected address
+	close(b.upc)
+	b.downc = make(chan struct{})
+	b.pinAddr = addr.Addr
+	// notify client that a connection is up
+	b.readyOnce.Do(func() { close(b.readyc) })
+	return func(err error) {
+		b.mu.Lock()
+		b.upc = make(chan struct{})
+		close(b.downc)
+		b.pinAddr = ""
+		b.mu.Unlock()
+	}
+}
+
+func (b *simpleBalancer) Get(ctx context.Context, opts grpc.BalancerGetOptions) (grpc.Address, func(), error) {
+	var (
+		addr   string
+		closed bool
+	)
+
+	// If opts.BlockingWait is false (for fail-fast RPCs), it should return
+	// an address it has notified via Notify immediately instead of blocking.
+	if !opts.BlockingWait {
+		b.mu.RLock()
+		closed = b.closed
+		addr = b.pinAddr
+		b.mu.RUnlock()
+		if closed {
+			return grpc.Address{Addr: ""}, nil, grpc.ErrClientConnClosing
+		}
+		if addr == "" {
+			return grpc.Address{Addr: ""}, nil, ErrNoAddrAvilable
+		}
+		return grpc.Address{Addr: addr}, func() {}, nil
+	}
+
+	for {
+		b.mu.RLock()
+		ch := b.upc
+		b.mu.RUnlock()
+		select {
+		case <-ch:
+		case <-b.donec:
+			return grpc.Address{Addr: ""}, nil, grpc.ErrClientConnClosing
+		case <-ctx.Done():
+			return grpc.Address{Addr: ""}, nil, ctx.Err()
+		}
+		b.mu.RLock()
+		closed = b.closed
+		addr = b.pinAddr
+		b.mu.RUnlock()
+		// Close() which sets b.closed = true can be called before Get(), Get() must exit if balancer is closed.
+		if closed {
+			return grpc.Address{Addr: ""}, nil, grpc.ErrClientConnClosing
+		}
+		if addr != "" {
+			break
+		}
+	}
+	return grpc.Address{Addr: addr}, func() {}, nil
+}
+
+func (b *simpleBalancer) Notify() <-chan []grpc.Address { return b.notifyCh }
+
+func (b *simpleBalancer) Close() error {
+	b.mu.Lock()
+	// In case gRPC calls close twice. TODO: remove the checking
+	// when we are sure that gRPC wont call close twice.
+	if b.closed {
+		b.mu.Unlock()
+		<-b.donec
+		return nil
+	}
+	b.closed = true
+	close(b.stopc)
+	b.pinAddr = ""
+
+	// In the case of following scenario:
+	//	1. upc is not closed; no pinned address
+	// 	2. client issues an rpc, calling invoke(), which calls Get(), enters for loop, blocks
+	// 	3. clientconn.Close() calls balancer.Close(); closed = true
+	// 	4. for loop in Get() never exits since ctx is the context passed in by the client and may not be canceled
+	// we must close upc so Get() exits from blocking on upc
+	select {
+	case <-b.upc:
+	default:
+		// terminate all waiting Get()s
+		close(b.upc)
+	}
+
+	b.mu.Unlock()
+
+	// wait for updateNotifyLoop to finish
+	<-b.donec
+	close(b.notifyCh)
+
+	return nil
+}
+
+func getHost(ep string) string {
+	url, uerr := url.Parse(ep)
+	if uerr != nil || !strings.Contains(ep, "://") {
+		return ep
+	}
+	return url.Host
+}

vendor/github.com/coreos/etcd/clientv3/client.go

@@ -0,0 +1,515 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"crypto/tls"
+	"errors"
+	"fmt"
+	"net"
+	"net/url"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes"
+
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+	"google.golang.org/grpc/credentials"
+	"google.golang.org/grpc/metadata"
+)
+
+var (
+	ErrNoAvailableEndpoints = errors.New("etcdclient: no available endpoints")
+	ErrOldCluster           = errors.New("etcdclient: old cluster version")
+)
+
+// Client provides and manages an etcd v3 client session.
+type Client struct {
+	Cluster
+	KV
+	Lease
+	Watcher
+	Auth
+	Maintenance
+
+	conn     *grpc.ClientConn
+	dialerrc chan error
+
+	cfg              Config
+	creds            *credentials.TransportCredentials
+	balancer         *simpleBalancer
+	retryWrapper     retryRpcFunc
+	retryAuthWrapper retryRpcFunc
+
+	ctx    context.Context
+	cancel context.CancelFunc
+
+	// Username is a username for authentication
+	Username string
+	// Password is a password for authentication
+	Password string
+	// tokenCred is an instance of WithPerRPCCredentials()'s argument
+	tokenCred *authTokenCredential
+}
+
+// New creates a new etcdv3 client from a given configuration.
+func New(cfg Config) (*Client, error) {
+	if len(cfg.Endpoints) == 0 {
+		return nil, ErrNoAvailableEndpoints
+	}
+
+	return newClient(&cfg)
+}
+
+// NewCtxClient creates a client with a context but no underlying grpc
+// connection. This is useful for embedded cases that override the
+// service interface implementations and do not need connection management.
+func NewCtxClient(ctx context.Context) *Client {
+	cctx, cancel := context.WithCancel(ctx)
+	return &Client{ctx: cctx, cancel: cancel}
+}
+
+// NewFromURL creates a new etcdv3 client from a URL.
+func NewFromURL(url string) (*Client, error) {
+	return New(Config{Endpoints: []string{url}})
+}
+
+// Close shuts down the client's etcd connections.
+func (c *Client) Close() error {
+	c.cancel()
+	c.Watcher.Close()
+	c.Lease.Close()
+	if c.conn != nil {
+		return toErr(c.ctx, c.conn.Close())
+	}
+	return c.ctx.Err()
+}
+
+// Ctx is a context for "out of band" messages (e.g., for sending
+// "clean up" message when another context is canceled). It is
+// canceled on client Close().
+func (c *Client) Ctx() context.Context { return c.ctx }
+
+// Endpoints lists the registered endpoints for the client.
+func (c *Client) Endpoints() (eps []string) {
+	// copy the slice; protect original endpoints from being changed
+	eps = make([]string, len(c.cfg.Endpoints))
+	copy(eps, c.cfg.Endpoints)
+	return
+}
+
+// SetEndpoints updates client's endpoints.
+func (c *Client) SetEndpoints(eps ...string) {
+	c.cfg.Endpoints = eps
+	c.balancer.updateAddrs(eps)
+}
+
+// Sync synchronizes client's endpoints with the known endpoints from the etcd membership.
+func (c *Client) Sync(ctx context.Context) error {
+	mresp, err := c.MemberList(ctx)
+	if err != nil {
+		return err
+	}
+	var eps []string
+	for _, m := range mresp.Members {
+		eps = append(eps, m.ClientURLs...)
+	}
+	c.SetEndpoints(eps...)
+	return nil
+}
+
+func (c *Client) autoSync() {
+	if c.cfg.AutoSyncInterval == time.Duration(0) {
+		return
+	}
+
+	for {
+		select {
+		case <-c.ctx.Done():
+			return
+		case <-time.After(c.cfg.AutoSyncInterval):
+			ctx, _ := context.WithTimeout(c.ctx, 5*time.Second)
+			if err := c.Sync(ctx); err != nil && err != c.ctx.Err() {
+				logger.Println("Auto sync endpoints failed:", err)
+			}
+		}
+	}
+}
+
+type authTokenCredential struct {
+	token   string
+	tokenMu *sync.RWMutex
+}
+
+func (cred authTokenCredential) RequireTransportSecurity() bool {
+	return false
+}
+
+func (cred authTokenCredential) GetRequestMetadata(ctx context.Context, s ...string) (map[string]string, error) {
+	cred.tokenMu.RLock()
+	defer cred.tokenMu.RUnlock()
+	return map[string]string{
+		"token": cred.token,
+	}, nil
+}
+
+func parseEndpoint(endpoint string) (proto string, host string, scheme string) {
+	proto = "tcp"
+	host = endpoint
+	url, uerr := url.Parse(endpoint)
+	if uerr != nil || !strings.Contains(endpoint, "://") {
+		return
+	}
+	scheme = url.Scheme
+
+	// strip scheme:// prefix since grpc dials by host
+	host = url.Host
+	switch url.Scheme {
+	case "http", "https":
+	case "unix", "unixs":
+		proto = "unix"
+		host = url.Host + url.Path
+	default:
+		proto, host = "", ""
+	}
+	return
+}
+
+func (c *Client) processCreds(scheme string) (creds *credentials.TransportCredentials) {
+	creds = c.creds
+	switch scheme {
+	case "unix":
+	case "http":
+		creds = nil
+	case "https", "unixs":
+		if creds != nil {
+			break
+		}
+		tlsconfig := &tls.Config{}
+		emptyCreds := credentials.NewTLS(tlsconfig)
+		creds = &emptyCreds
+	default:
+		creds = nil
+	}
+	return
+}
+
+// dialSetupOpts gives the dial opts prior to any authentication
+func (c *Client) dialSetupOpts(endpoint string, dopts ...grpc.DialOption) (opts []grpc.DialOption) {
+	if c.cfg.DialTimeout > 0 {
+		opts = []grpc.DialOption{grpc.WithTimeout(c.cfg.DialTimeout)}
+	}
+	opts = append(opts, dopts...)
+
+	f := func(host string, t time.Duration) (net.Conn, error) {
+		proto, host, _ := parseEndpoint(c.balancer.getEndpoint(host))
+		if host == "" && endpoint != "" {
+			// dialing an endpoint not in the balancer; use
+			// endpoint passed into dial
+			proto, host, _ = parseEndpoint(endpoint)
+		}
+		if proto == "" {
+			return nil, fmt.Errorf("unknown scheme for %q", host)
+		}
+		select {
+		case <-c.ctx.Done():
+			return nil, c.ctx.Err()
+		default:
+		}
+		dialer := &net.Dialer{Timeout: t}
+		conn, err := dialer.DialContext(c.ctx, proto, host)
+		if err != nil {
+			select {
+			case c.dialerrc <- err:
+			default:
+			}
+		}
+		return conn, err
+	}
+	opts = append(opts, grpc.WithDialer(f))
+
+	creds := c.creds
+	if _, _, scheme := parseEndpoint(endpoint); len(scheme) != 0 {
+		creds = c.processCreds(scheme)
+	}
+	if creds != nil {
+		opts = append(opts, grpc.WithTransportCredentials(*creds))
+	} else {
+		opts = append(opts, grpc.WithInsecure())
+	}
+
+	return opts
+}
+
+// Dial connects to a single endpoint using the client's config.
+func (c *Client) Dial(endpoint string) (*grpc.ClientConn, error) {
+	return c.dial(endpoint)
+}
+
+func (c *Client) getToken(ctx context.Context) error {
+	var err error // return last error in a case of fail
+	var auth *authenticator
+
+	for i := 0; i < len(c.cfg.Endpoints); i++ {
+		endpoint := c.cfg.Endpoints[i]
+		host := getHost(endpoint)
+		// use dial options without dopts to avoid reusing the client balancer
+		auth, err = newAuthenticator(host, c.dialSetupOpts(endpoint))
+		if err != nil {
+			continue
+		}
+		defer auth.close()
+
+		var resp *AuthenticateResponse
+		resp, err = auth.authenticate(ctx, c.Username, c.Password)
+		if err != nil {
+			continue
+		}
+
+		c.tokenCred.tokenMu.Lock()
+		c.tokenCred.token = resp.Token
+		c.tokenCred.tokenMu.Unlock()
+
+		return nil
+	}
+
+	return err
+}
+
+func (c *Client) dial(endpoint string, dopts ...grpc.DialOption) (*grpc.ClientConn, error) {
+	opts := c.dialSetupOpts(endpoint, dopts...)
+	host := getHost(endpoint)
+	if c.Username != "" && c.Password != "" {
+		c.tokenCred = &authTokenCredential{
+			tokenMu: &sync.RWMutex{},
+		}
+
+		ctx := c.ctx
+		if c.cfg.DialTimeout > 0 {
+			cctx, cancel := context.WithTimeout(ctx, c.cfg.DialTimeout)
+			defer cancel()
+			ctx = cctx
+		}
+
+		err := c.getToken(ctx)
+		if err != nil {
+			if toErr(ctx, err) != rpctypes.ErrAuthNotEnabled {
+				if err == ctx.Err() && ctx.Err() != c.ctx.Err() {
+					err = grpc.ErrClientConnTimeout
+				}
+				return nil, err
+			}
+		} else {
+			opts = append(opts, grpc.WithPerRPCCredentials(c.tokenCred))
+		}
+	}
+
+	opts = append(opts, c.cfg.DialOptions...)
+
+	conn, err := grpc.DialContext(c.ctx, host, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return conn, nil
+}
+
+// WithRequireLeader requires client requests to only succeed
+// when the cluster has a leader.
+func WithRequireLeader(ctx context.Context) context.Context {
+	md := metadata.Pairs(rpctypes.MetadataRequireLeaderKey, rpctypes.MetadataHasLeader)
+	return metadata.NewContext(ctx, md)
+}
+
+func newClient(cfg *Config) (*Client, error) {
+	if cfg == nil {
+		cfg = &Config{}
+	}
+	var creds *credentials.TransportCredentials
+	if cfg.TLS != nil {
+		c := credentials.NewTLS(cfg.TLS)
+		creds = &c
+	}
+
+	// use a temporary skeleton client to bootstrap first connection
+	baseCtx := context.TODO()
+	if cfg.Context != nil {
+		baseCtx = cfg.Context
+	}
+
+	ctx, cancel := context.WithCancel(baseCtx)
+	client := &Client{
+		conn:     nil,
+		dialerrc: make(chan error, 1),
+		cfg:      *cfg,
+		creds:    creds,
+		ctx:      ctx,
+		cancel:   cancel,
+	}
+	if cfg.Username != "" && cfg.Password != "" {
+		client.Username = cfg.Username
+		client.Password = cfg.Password
+	}
+
+	client.balancer = newSimpleBalancer(cfg.Endpoints)
+	// use Endpoints[0] so that for https:// without any tls config given, then
+	// grpc will assume the ServerName is in the endpoint.
+	conn, err := client.dial(cfg.Endpoints[0], grpc.WithBalancer(client.balancer))
+	if err != nil {
+		client.cancel()
+		client.balancer.Close()
+		return nil, err
+	}
+	client.conn = conn
+	client.retryWrapper = client.newRetryWrapper()
+	client.retryAuthWrapper = client.newAuthRetryWrapper()
+
+	// wait for a connection
+	if cfg.DialTimeout > 0 {
+		hasConn := false
+		waitc := time.After(cfg.DialTimeout)
+		select {
+		case <-client.balancer.readyc:
+			hasConn = true
+		case <-ctx.Done():
+		case <-waitc:
+		}
+		if !hasConn {
+			err := grpc.ErrClientConnTimeout
+			select {
+			case err = <-client.dialerrc:
+			default:
+			}
+			client.cancel()
+			client.balancer.Close()
+			conn.Close()
+			return nil, err
+		}
+	}
+
+	client.Cluster = NewCluster(client)
+	client.KV = NewKV(client)
+	client.Lease = NewLease(client)
+	client.Watcher = NewWatcher(client)
+	client.Auth = NewAuth(client)
+	client.Maintenance = NewMaintenance(client)
+
+	if cfg.RejectOldCluster {
+		if err := client.checkVersion(); err != nil {
+			client.Close()
+			return nil, err
+		}
+	}
+
+	go client.autoSync()
+	return client, nil
+}
+
+func (c *Client) checkVersion() (err error) {
+	var wg sync.WaitGroup
+	errc := make(chan error, len(c.cfg.Endpoints))
+	ctx, cancel := context.WithCancel(c.ctx)
+	if c.cfg.DialTimeout > 0 {
+		ctx, _ = context.WithTimeout(ctx, c.cfg.DialTimeout)
+	}
+	wg.Add(len(c.cfg.Endpoints))
+	for _, ep := range c.cfg.Endpoints {
+		// if cluster is current, any endpoint gives a recent version
+		go func(e string) {
+			defer wg.Done()
+			resp, rerr := c.Status(ctx, e)
+			if rerr != nil {
+				errc <- rerr
+				return
+			}
+			vs := strings.Split(resp.Version, ".")
+			maj, min := 0, 0
+			if len(vs) >= 2 {
+				maj, rerr = strconv.Atoi(vs[0])
+				min, rerr = strconv.Atoi(vs[1])
+			}
+			if maj < 3 || (maj == 3 && min < 2) {
+				rerr = ErrOldCluster
+			}
+			errc <- rerr
+		}(ep)
+	}
+	// wait for success
+	for i := 0; i < len(c.cfg.Endpoints); i++ {
+		if err = <-errc; err == nil {
+			break
+		}
+	}
+	cancel()
+	wg.Wait()
+	return err
+}
+
+// ActiveConnection returns the current in-use connection
+func (c *Client) ActiveConnection() *grpc.ClientConn { return c.conn }
+
+// isHaltErr returns true if the given error and context indicate no forward
+// progress can be made, even after reconnecting.
+func isHaltErr(ctx context.Context, err error) bool {
+	if ctx != nil && ctx.Err() != nil {
+		return true
+	}
+	if err == nil {
+		return false
+	}
+	code := grpc.Code(err)
+	// Unavailable codes mean the system will be right back.
+	// (e.g., can't connect, lost leader)
+	// Treat Internal codes as if something failed, leaving the
+	// system in an inconsistent state, but retrying could make progress.
+	// (e.g., failed in middle of send, corrupted frame)
+	// TODO: are permanent Internal errors possible from grpc?
+	return code != codes.Unavailable && code != codes.Internal
+}
+
+func toErr(ctx context.Context, err error) error {
+	if err == nil {
+		return nil
+	}
+	err = rpctypes.Error(err)
+	if _, ok := err.(rpctypes.EtcdError); ok {
+		return err
+	}
+	code := grpc.Code(err)
+	switch code {
+	case codes.DeadlineExceeded:
+		fallthrough
+	case codes.Canceled:
+		if ctx.Err() != nil {
+			err = ctx.Err()
+		}
+	case codes.Unavailable:
+		err = ErrNoAvailableEndpoints
+	case codes.FailedPrecondition:
+		err = grpc.ErrClientConnClosing
+	}
+	return err
+}
+
+func canceledByCaller(stopCtx context.Context, err error) bool {
+	if stopCtx.Err() == nil || err == nil {
+		return false
+	}
+
+	return err == context.Canceled || err == context.DeadlineExceeded
+}

vendor/github.com/coreos/etcd/clientv3/cluster.go

@@ -0,0 +1,100 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+)
+
+type (
+	Member               pb.Member
+	MemberListResponse   pb.MemberListResponse
+	MemberAddResponse    pb.MemberAddResponse
+	MemberRemoveResponse pb.MemberRemoveResponse
+	MemberUpdateResponse pb.MemberUpdateResponse
+)
+
+type Cluster interface {
+	// MemberList lists the current cluster membership.
+	MemberList(ctx context.Context) (*MemberListResponse, error)
+
+	// MemberAdd adds a new member into the cluster.
+	MemberAdd(ctx context.Context, peerAddrs []string) (*MemberAddResponse, error)
+
+	// MemberRemove removes an existing member from the cluster.
+	MemberRemove(ctx context.Context, id uint64) (*MemberRemoveResponse, error)
+
+	// MemberUpdate updates the peer addresses of the member.
+	MemberUpdate(ctx context.Context, id uint64, peerAddrs []string) (*MemberUpdateResponse, error)
+}
+
+type cluster struct {
+	remote pb.ClusterClient
+}
+
+func NewCluster(c *Client) Cluster {
+	return &cluster{remote: RetryClusterClient(c)}
+}
+
+func NewClusterFromClusterClient(remote pb.ClusterClient) Cluster {
+	return &cluster{remote: remote}
+}
+
+func (c *cluster) MemberAdd(ctx context.Context, peerAddrs []string) (*MemberAddResponse, error) {
+	r := &pb.MemberAddRequest{PeerURLs: peerAddrs}
+	resp, err := c.remote.MemberAdd(ctx, r)
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	return (*MemberAddResponse)(resp), nil
+}
+
+func (c *cluster) MemberRemove(ctx context.Context, id uint64) (*MemberRemoveResponse, error) {
+	r := &pb.MemberRemoveRequest{ID: id}
+	resp, err := c.remote.MemberRemove(ctx, r)
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	return (*MemberRemoveResponse)(resp), nil
+}
+
+func (c *cluster) MemberUpdate(ctx context.Context, id uint64, peerAddrs []string) (*MemberUpdateResponse, error) {
+	// it is safe to retry on update.
+	for {
+		r := &pb.MemberUpdateRequest{ID: id, PeerURLs: peerAddrs}
+		resp, err := c.remote.MemberUpdate(ctx, r, grpc.FailFast(false))
+		if err == nil {
+			return (*MemberUpdateResponse)(resp), nil
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}
+
+func (c *cluster) MemberList(ctx context.Context) (*MemberListResponse, error) {
+	// it is safe to retry on list.
+	for {
+		resp, err := c.remote.MemberList(ctx, &pb.MemberListRequest{}, grpc.FailFast(false))
+		if err == nil {
+			return (*MemberListResponse)(resp), nil
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}

vendor/github.com/coreos/etcd/clientv3/compact_op.go

@@ -0,0 +1,53 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+)
+
+// CompactOp represents a compact operation.
+type CompactOp struct {
+	revision int64
+	physical bool
+}
+
+// CompactOption configures compact operation.
+type CompactOption func(*CompactOp)
+
+func (op *CompactOp) applyCompactOpts(opts []CompactOption) {
+	for _, opt := range opts {
+		opt(op)
+	}
+}
+
+// OpCompact wraps slice CompactOption to create a CompactOp.
+func OpCompact(rev int64, opts ...CompactOption) CompactOp {
+	ret := CompactOp{revision: rev}
+	ret.applyCompactOpts(opts)
+	return ret
+}
+
+func (op CompactOp) toRequest() *pb.CompactionRequest {
+	return &pb.CompactionRequest{Revision: op.revision, Physical: op.physical}
+}
+
+// WithCompactPhysical makes compact RPC call wait until
+// the compaction is physically applied to the local database
+// such that compacted entries are totally removed from the
+// backend database.
+func WithCompactPhysical() CompactOption {
+	return func(op *CompactOp) { op.physical = true }
+}

vendor/github.com/coreos/etcd/clientv3/compare.go

@@ -0,0 +1,110 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+)
+
+type CompareTarget int
+type CompareResult int
+
+const (
+	CompareVersion CompareTarget = iota
+	CompareCreated
+	CompareModified
+	CompareValue
+)
+
+type Cmp pb.Compare
+
+func Compare(cmp Cmp, result string, v interface{}) Cmp {
+	var r pb.Compare_CompareResult
+
+	switch result {
+	case "=":
+		r = pb.Compare_EQUAL
+	case "!=":
+		r = pb.Compare_NOT_EQUAL
+	case ">":
+		r = pb.Compare_GREATER
+	case "<":
+		r = pb.Compare_LESS
+	default:
+		panic("Unknown result op")
+	}
+
+	cmp.Result = r
+	switch cmp.Target {
+	case pb.Compare_VALUE:
+		val, ok := v.(string)
+		if !ok {
+			panic("bad compare value")
+		}
+		cmp.TargetUnion = &pb.Compare_Value{Value: []byte(val)}
+	case pb.Compare_VERSION:
+		cmp.TargetUnion = &pb.Compare_Version{Version: mustInt64(v)}
+	case pb.Compare_CREATE:
+		cmp.TargetUnion = &pb.Compare_CreateRevision{CreateRevision: mustInt64(v)}
+	case pb.Compare_MOD:
+		cmp.TargetUnion = &pb.Compare_ModRevision{ModRevision: mustInt64(v)}
+	default:
+		panic("Unknown compare type")
+	}
+	return cmp
+}
+
+func Value(key string) Cmp {
+	return Cmp{Key: []byte(key), Target: pb.Compare_VALUE}
+}
+
+func Version(key string) Cmp {
+	return Cmp{Key: []byte(key), Target: pb.Compare_VERSION}
+}
+
+func CreateRevision(key string) Cmp {
+	return Cmp{Key: []byte(key), Target: pb.Compare_CREATE}
+}
+
+func ModRevision(key string) Cmp {
+	return Cmp{Key: []byte(key), Target: pb.Compare_MOD}
+}
+
+// KeyBytes returns the byte slice holding with the comparison key.
+func (cmp *Cmp) KeyBytes() []byte { return cmp.Key }
+
+// WithKeyBytes sets the byte slice for the comparison key.
+func (cmp *Cmp) WithKeyBytes(key []byte) { cmp.Key = key }
+
+// ValueBytes returns the byte slice holding the comparison value, if any.
+func (cmp *Cmp) ValueBytes() []byte {
+	if tu, ok := cmp.TargetUnion.(*pb.Compare_Value); ok {
+		return tu.Value
+	}
+	return nil
+}
+
+// WithValueBytes sets the byte slice for the comparison's value.
+func (cmp *Cmp) WithValueBytes(v []byte) { cmp.TargetUnion.(*pb.Compare_Value).Value = v }
+
+func mustInt64(val interface{}) int64 {
+	if v, ok := val.(int64); ok {
+		return v
+	}
+	if v, ok := val.(int); ok {
+		return int64(v)
+	}
+	panic("bad value")
+}

vendor/github.com/coreos/etcd/clientv3/concurrency/doc.go

@@ -0,0 +1,17 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Package concurrency implements concurrency operations on top of
+// etcd such as distributed locks, barriers, and elections.
+package concurrency

vendor/github.com/coreos/etcd/clientv3/concurrency/election.go

@@ -0,0 +1,243 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package concurrency
+
+import (
+	"errors"
+	"fmt"
+
+	v3 "github.com/coreos/etcd/clientv3"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"github.com/coreos/etcd/mvcc/mvccpb"
+	"golang.org/x/net/context"
+)
+
+var (
+	ErrElectionNotLeader = errors.New("election: not leader")
+	ErrElectionNoLeader  = errors.New("election: no leader")
+)
+
+type Election struct {
+	session *Session
+
+	keyPrefix string
+
+	leaderKey     string
+	leaderRev     int64
+	leaderSession *Session
+	hdr           *pb.ResponseHeader
+}
+
+// NewElection returns a new election on a given key prefix.
+func NewElection(s *Session, pfx string) *Election {
+	return &Election{session: s, keyPrefix: pfx + "/"}
+}
+
+// ResumeElection initializes an election with a known leader.
+func ResumeElection(s *Session, pfx string, leaderKey string, leaderRev int64) *Election {
+	return &Election{
+		session:       s,
+		leaderKey:     leaderKey,
+		leaderRev:     leaderRev,
+		leaderSession: s,
+	}
+}
+
+// Campaign puts a value as eligible for the election. It blocks until
+// it is elected, an error occurs, or the context is cancelled.
+func (e *Election) Campaign(ctx context.Context, val string) error {
+	s := e.session
+	client := e.session.Client()
+
+	k := fmt.Sprintf("%s%x", e.keyPrefix, s.Lease())
+	txn := client.Txn(ctx).If(v3.Compare(v3.CreateRevision(k), "=", 0))
+	txn = txn.Then(v3.OpPut(k, val, v3.WithLease(s.Lease())))
+	txn = txn.Else(v3.OpGet(k))
+	resp, err := txn.Commit()
+	if err != nil {
+		return err
+	}
+	e.leaderKey, e.leaderRev, e.leaderSession = k, resp.Header.Revision, s
+	if !resp.Succeeded {
+		kv := resp.Responses[0].GetResponseRange().Kvs[0]
+		e.leaderRev = kv.CreateRevision
+		if string(kv.Value) != val {
+			if err = e.Proclaim(ctx, val); err != nil {
+				e.Resign(ctx)
+				return err
+			}
+		}
+	}
+
+	_, err = waitDeletes(ctx, client, e.keyPrefix, e.leaderRev-1)
+	if err != nil {
+		// clean up in case of context cancel
+		select {
+		case <-ctx.Done():
+			e.Resign(client.Ctx())
+		default:
+			e.leaderSession = nil
+		}
+		return err
+	}
+	e.hdr = resp.Header
+
+	return nil
+}
+
+// Proclaim lets the leader announce a new value without another election.
+func (e *Election) Proclaim(ctx context.Context, val string) error {
+	if e.leaderSession == nil {
+		return ErrElectionNotLeader
+	}
+	client := e.session.Client()
+	cmp := v3.Compare(v3.CreateRevision(e.leaderKey), "=", e.leaderRev)
+	txn := client.Txn(ctx).If(cmp)
+	txn = txn.Then(v3.OpPut(e.leaderKey, val, v3.WithLease(e.leaderSession.Lease())))
+	tresp, terr := txn.Commit()
+	if terr != nil {
+		return terr
+	}
+	if !tresp.Succeeded {
+		e.leaderKey = ""
+		return ErrElectionNotLeader
+	}
+
+	e.hdr = tresp.Header
+	return nil
+}
+
+// Resign lets a leader start a new election.
+func (e *Election) Resign(ctx context.Context) (err error) {
+	if e.leaderSession == nil {
+		return nil
+	}
+	client := e.session.Client()
+	cmp := v3.Compare(v3.CreateRevision(e.leaderKey), "=", e.leaderRev)
+	resp, err := client.Txn(ctx).If(cmp).Then(v3.OpDelete(e.leaderKey)).Commit()
+	if err == nil {
+		e.hdr = resp.Header
+	}
+	e.leaderKey = ""
+	e.leaderSession = nil
+	return err
+}
+
+// Leader returns the leader value for the current election.
+func (e *Election) Leader(ctx context.Context) (*v3.GetResponse, error) {
+	client := e.session.Client()
+	resp, err := client.Get(ctx, e.keyPrefix, v3.WithFirstCreate()...)
+	if err != nil {
+		return nil, err
+	} else if len(resp.Kvs) == 0 {
+		// no leader currently elected
+		return nil, ErrElectionNoLeader
+	}
+	return resp, nil
+}
+
+// Observe returns a channel that reliably observes ordered leader proposals
+// as GetResponse values on every current elected leader key. It will not
+// necessarily fetch all historical leader updates, but will always post the
+// most recent leader value.
+//
+// The channel closes when the context is canceled or the underlying watcher
+// is otherwise disrupted.
+func (e *Election) Observe(ctx context.Context) <-chan v3.GetResponse {
+	retc := make(chan v3.GetResponse)
+	go e.observe(ctx, retc)
+	return retc
+}
+
+func (e *Election) observe(ctx context.Context, ch chan<- v3.GetResponse) {
+	client := e.session.Client()
+
+	defer close(ch)
+	for {
+		resp, err := client.Get(ctx, e.keyPrefix, v3.WithFirstCreate()...)
+		if err != nil {
+			return
+		}
+
+		var kv *mvccpb.KeyValue
+		var hdr *pb.ResponseHeader
+
+		if len(resp.Kvs) == 0 {
+			cctx, cancel := context.WithCancel(ctx)
+			// wait for first key put on prefix
+			opts := []v3.OpOption{v3.WithRev(resp.Header.Revision), v3.WithPrefix()}
+			wch := client.Watch(cctx, e.keyPrefix, opts...)
+			for kv == nil {
+				wr, ok := <-wch
+				if !ok || wr.Err() != nil {
+					cancel()
+					return
+				}
+				// only accept PUTs; a DELETE will make observe() spin
+				for _, ev := range wr.Events {
+					if ev.Type == mvccpb.PUT {
+						hdr, kv = &wr.Header, ev.Kv
+						// may have multiple revs; hdr.rev = the last rev
+						// set to kv's rev in case batch has multiple PUTs
+						hdr.Revision = kv.ModRevision
+						break
+					}
+				}
+			}
+			cancel()
+		} else {
+			hdr, kv = resp.Header, resp.Kvs[0]
+		}
+
+		select {
+		case ch <- v3.GetResponse{Header: hdr, Kvs: []*mvccpb.KeyValue{kv}}:
+		case <-ctx.Done():
+			return
+		}
+
+		cctx, cancel := context.WithCancel(ctx)
+		wch := client.Watch(cctx, string(kv.Key), v3.WithRev(hdr.Revision+1))
+		keyDeleted := false
+		for !keyDeleted {
+			wr, ok := <-wch
+			if !ok {
+				return
+			}
+			for _, ev := range wr.Events {
+				if ev.Type == mvccpb.DELETE {
+					keyDeleted = true
+					break
+				}
+				resp.Header = &wr.Header
+				resp.Kvs = []*mvccpb.KeyValue{ev.Kv}
+				select {
+				case ch <- *resp:
+				case <-cctx.Done():
+					return
+				}
+			}
+		}
+		cancel()
+	}
+}
+
+// Key returns the leader key if elected, empty string otherwise.
+func (e *Election) Key() string { return e.leaderKey }
+
+// Rev returns the leader key's creation revision, if elected.
+func (e *Election) Rev() int64 { return e.leaderRev }
+
+// Header is the response header from the last successful election proposal.
+func (m *Election) Header() *pb.ResponseHeader { return m.hdr }

vendor/github.com/coreos/etcd/clientv3/concurrency/key.go

@@ -0,0 +1,65 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package concurrency
+
+import (
+	"fmt"
+
+	v3 "github.com/coreos/etcd/clientv3"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"github.com/coreos/etcd/mvcc/mvccpb"
+	"golang.org/x/net/context"
+)
+
+func waitDelete(ctx context.Context, client *v3.Client, key string, rev int64) error {
+	cctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+
+	var wr v3.WatchResponse
+	wch := client.Watch(cctx, key, v3.WithRev(rev))
+	for wr = range wch {
+		for _, ev := range wr.Events {
+			if ev.Type == mvccpb.DELETE {
+				return nil
+			}
+		}
+	}
+	if err := wr.Err(); err != nil {
+		return err
+	}
+	if err := ctx.Err(); err != nil {
+		return err
+	}
+	return fmt.Errorf("lost watcher waiting for delete")
+}
+
+// waitDeletes efficiently waits until all keys matching the prefix and no greater
+// than the create revision.
+func waitDeletes(ctx context.Context, client *v3.Client, pfx string, maxCreateRev int64) (*pb.ResponseHeader, error) {
+	getOpts := append(v3.WithLastCreate(), v3.WithMaxCreateRev(maxCreateRev))
+	for {
+		resp, err := client.Get(ctx, pfx, getOpts...)
+		if err != nil {
+			return nil, err
+		}
+		if len(resp.Kvs) == 0 {
+			return resp.Header, nil
+		}
+		lastKey := string(resp.Kvs[0].Key)
+		if err = waitDelete(ctx, client, lastKey, resp.Header.Revision); err != nil {
+			return nil, err
+		}
+	}
+}

vendor/github.com/coreos/etcd/clientv3/concurrency/mutex.go

@@ -0,0 +1,110 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package concurrency
+
+import (
+	"fmt"
+	"sync"
+
+	v3 "github.com/coreos/etcd/clientv3"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+)
+
+// Mutex implements the sync Locker interface with etcd
+type Mutex struct {
+	s *Session
+
+	pfx   string
+	myKey string
+	myRev int64
+	hdr   *pb.ResponseHeader
+}
+
+func NewMutex(s *Session, pfx string) *Mutex {
+	return &Mutex{s, pfx + "/", "", -1, nil}
+}
+
+// Lock locks the mutex with a cancelable context. If the context is canceled
+// while trying to acquire the lock, the mutex tries to clean its stale lock entry.
+func (m *Mutex) Lock(ctx context.Context) error {
+	s := m.s
+	client := m.s.Client()
+
+	m.myKey = fmt.Sprintf("%s%x", m.pfx, s.Lease())
+	cmp := v3.Compare(v3.CreateRevision(m.myKey), "=", 0)
+	// put self in lock waiters via myKey; oldest waiter holds lock
+	put := v3.OpPut(m.myKey, "", v3.WithLease(s.Lease()))
+	// reuse key in case this session already holds the lock
+	get := v3.OpGet(m.myKey)
+	resp, err := client.Txn(ctx).If(cmp).Then(put).Else(get).Commit()
+	if err != nil {
+		return err
+	}
+	m.myRev = resp.Header.Revision
+	if !resp.Succeeded {
+		m.myRev = resp.Responses[0].GetResponseRange().Kvs[0].CreateRevision
+	}
+
+	// wait for deletion revisions prior to myKey
+	hdr, werr := waitDeletes(ctx, client, m.pfx, m.myRev-1)
+	// release lock key if cancelled
+	select {
+	case <-ctx.Done():
+		m.Unlock(client.Ctx())
+	default:
+		m.hdr = hdr
+	}
+	return werr
+}
+
+func (m *Mutex) Unlock(ctx context.Context) error {
+	client := m.s.Client()
+	if _, err := client.Delete(ctx, m.myKey); err != nil {
+		return err
+	}
+	m.myKey = "\x00"
+	m.myRev = -1
+	return nil
+}
+
+func (m *Mutex) IsOwner() v3.Cmp {
+	return v3.Compare(v3.CreateRevision(m.myKey), "=", m.myRev)
+}
+
+func (m *Mutex) Key() string { return m.myKey }
+
+// Header is the response header received from etcd on acquiring the lock.
+func (m *Mutex) Header() *pb.ResponseHeader { return m.hdr }
+
+type lockerMutex struct{ *Mutex }
+
+func (lm *lockerMutex) Lock() {
+	client := lm.s.Client()
+	if err := lm.Mutex.Lock(client.Ctx()); err != nil {
+		panic(err)
+	}
+}
+func (lm *lockerMutex) Unlock() {
+	client := lm.s.Client()
+	if err := lm.Mutex.Unlock(client.Ctx()); err != nil {
+		panic(err)
+	}
+}
+
+// NewLocker creates a sync.Locker backed by an etcd mutex.
+func NewLocker(s *Session, pfx string) sync.Locker {
+	return &lockerMutex{NewMutex(s, pfx)}
+}

vendor/github.com/coreos/etcd/clientv3/concurrency/session.go

@@ -0,0 +1,140 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package concurrency
+
+import (
+	"time"
+
+	v3 "github.com/coreos/etcd/clientv3"
+	"golang.org/x/net/context"
+)
+
+const defaultSessionTTL = 60
+
+// Session represents a lease kept alive for the lifetime of a client.
+// Fault-tolerant applications may use sessions to reason about liveness.
+type Session struct {
+	client *v3.Client
+	opts   *sessionOptions
+	id     v3.LeaseID
+
+	cancel context.CancelFunc
+	donec  <-chan struct{}
+}
+
+// NewSession gets the leased session for a client.
+func NewSession(client *v3.Client, opts ...SessionOption) (*Session, error) {
+	ops := &sessionOptions{ttl: defaultSessionTTL, ctx: client.Ctx()}
+	for _, opt := range opts {
+		opt(ops)
+	}
+
+	id := ops.leaseID
+	if id == v3.NoLease {
+		resp, err := client.Grant(ops.ctx, int64(ops.ttl))
+		if err != nil {
+			return nil, err
+		}
+		id = v3.LeaseID(resp.ID)
+	}
+
+	ctx, cancel := context.WithCancel(ops.ctx)
+	keepAlive, err := client.KeepAlive(ctx, id)
+	if err != nil || keepAlive == nil {
+		return nil, err
+	}
+
+	donec := make(chan struct{})
+	s := &Session{client: client, opts: ops, id: id, cancel: cancel, donec: donec}
+
+	// keep the lease alive until client error or cancelled context
+	go func() {
+		defer close(donec)
+		for range keepAlive {
+			// eat messages until keep alive channel closes
+		}
+	}()
+
+	return s, nil
+}
+
+// Client is the etcd client that is attached to the session.
+func (s *Session) Client() *v3.Client {
+	return s.client
+}
+
+// Lease is the lease ID for keys bound to the session.
+func (s *Session) Lease() v3.LeaseID { return s.id }
+
+// Done returns a channel that closes when the lease is orphaned, expires, or
+// is otherwise no longer being refreshed.
+func (s *Session) Done() <-chan struct{} { return s.donec }
+
+// Orphan ends the refresh for the session lease. This is useful
+// in case the state of the client connection is indeterminate (revoke
+// would fail) or when transferring lease ownership.
+func (s *Session) Orphan() {
+	s.cancel()
+	<-s.donec
+}
+
+// Close orphans the session and revokes the session lease.
+func (s *Session) Close() error {
+	s.Orphan()
+	// if revoke takes longer than the ttl, lease is expired anyway
+	ctx, cancel := context.WithTimeout(s.opts.ctx, time.Duration(s.opts.ttl)*time.Second)
+	_, err := s.client.Revoke(ctx, s.id)
+	cancel()
+	return err
+}
+
+type sessionOptions struct {
+	ttl     int
+	leaseID v3.LeaseID
+	ctx     context.Context
+}
+
+// SessionOption configures Session.
+type SessionOption func(*sessionOptions)
+
+// WithTTL configures the session's TTL in seconds.
+// If TTL is <= 0, the default 60 seconds TTL will be used.
+func WithTTL(ttl int) SessionOption {
+	return func(so *sessionOptions) {
+		if ttl > 0 {
+			so.ttl = ttl
+		}
+	}
+}
+
+// WithLease specifies the existing leaseID to be used for the session.
+// This is useful in process restart scenario, for example, to reclaim
+// leadership from an election prior to restart.
+func WithLease(leaseID v3.LeaseID) SessionOption {
+	return func(so *sessionOptions) {
+		so.leaseID = leaseID
+	}
+}
+
+// WithContext assigns a context to the session instead of defaulting to
+// using the client context. This is useful for canceling NewSession and
+// Close operations immediately without having to close the client. If the
+// context is canceled before Close() completes, the session's lease will be
+// abandoned and left to expire instead of being revoked.
+func WithContext(ctx context.Context) SessionOption {
+	return func(so *sessionOptions) {
+		so.ctx = ctx
+	}
+}

vendor/github.com/coreos/etcd/clientv3/concurrency/stm.go

@@ -0,0 +1,387 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package concurrency
+
+import (
+	"math"
+
+	v3 "github.com/coreos/etcd/clientv3"
+	"golang.org/x/net/context"
+)
+
+// STM is an interface for software transactional memory.
+type STM interface {
+	// Get returns the value for a key and inserts the key in the txn's read set.
+	// If Get fails, it aborts the transaction with an error, never returning.
+	Get(key ...string) string
+	// Put adds a value for a key to the write set.
+	Put(key, val string, opts ...v3.OpOption)
+	// Rev returns the revision of a key in the read set.
+	Rev(key string) int64
+	// Del deletes a key.
+	Del(key string)
+
+	// commit attempts to apply the txn's changes to the server.
+	commit() *v3.TxnResponse
+	reset()
+}
+
+// Isolation is an enumeration of transactional isolation levels which
+// describes how transactions should interfere and conflict.
+type Isolation int
+
+const (
+	// SerializableSnapshot provides serializable isolation and also checks
+	// for write conflicts.
+	SerializableSnapshot Isolation = iota
+	// Serializable reads within the same transactiona attempt return data
+	// from the at the revision of the first read.
+	Serializable
+	// RepeatableReads reads within the same transaction attempt always
+	// return the same data.
+	RepeatableReads
+	// ReadCommitted reads keys from any committed revision.
+	ReadCommitted
+)
+
+// stmError safely passes STM errors through panic to the STM error channel.
+type stmError struct{ err error }
+
+type stmOptions struct {
+	iso      Isolation
+	ctx      context.Context
+	prefetch []string
+}
+
+type stmOption func(*stmOptions)
+
+// WithIsolation specifies the transaction isolation level.
+func WithIsolation(lvl Isolation) stmOption {
+	return func(so *stmOptions) { so.iso = lvl }
+}
+
+// WithAbortContext specifies the context for permanently aborting the transaction.
+func WithAbortContext(ctx context.Context) stmOption {
+	return func(so *stmOptions) { so.ctx = ctx }
+}
+
+// WithPrefetch is a hint to prefetch a list of keys before trying to apply.
+// If an STM transaction will unconditionally fetch a set of keys, prefetching
+// those keys will save the round-trip cost from requesting each key one by one
+// with Get().
+func WithPrefetch(keys ...string) stmOption {
+	return func(so *stmOptions) { so.prefetch = append(so.prefetch, keys...) }
+}
+
+// NewSTM initiates a new STM instance, using snapshot isolation by default.
+func NewSTM(c *v3.Client, apply func(STM) error, so ...stmOption) (*v3.TxnResponse, error) {
+	opts := &stmOptions{ctx: c.Ctx()}
+	for _, f := range so {
+		f(opts)
+	}
+	if len(opts.prefetch) != 0 {
+		f := apply
+		apply = func(s STM) error {
+			s.Get(opts.prefetch...)
+			return f(s)
+		}
+	}
+	return runSTM(mkSTM(c, opts), apply)
+}
+
+func mkSTM(c *v3.Client, opts *stmOptions) STM {
+	switch opts.iso {
+	case SerializableSnapshot:
+		s := &stmSerializable{
+			stm:      stm{client: c, ctx: opts.ctx},
+			prefetch: make(map[string]*v3.GetResponse),
+		}
+		s.conflicts = func() []v3.Cmp {
+			return append(s.rset.cmps(), s.wset.cmps(s.rset.first()+1)...)
+		}
+		return s
+	case Serializable:
+		s := &stmSerializable{
+			stm:      stm{client: c, ctx: opts.ctx},
+			prefetch: make(map[string]*v3.GetResponse),
+		}
+		s.conflicts = func() []v3.Cmp { return s.rset.cmps() }
+		return s
+	case RepeatableReads:
+		s := &stm{client: c, ctx: opts.ctx, getOpts: []v3.OpOption{v3.WithSerializable()}}
+		s.conflicts = func() []v3.Cmp { return s.rset.cmps() }
+		return s
+	case ReadCommitted:
+		s := &stm{client: c, ctx: opts.ctx, getOpts: []v3.OpOption{v3.WithSerializable()}}
+		s.conflicts = func() []v3.Cmp { return nil }
+		return s
+	default:
+		panic("unsupported stm")
+	}
+}
+
+type stmResponse struct {
+	resp *v3.TxnResponse
+	err  error
+}
+
+func runSTM(s STM, apply func(STM) error) (*v3.TxnResponse, error) {
+	outc := make(chan stmResponse, 1)
+	go func() {
+		defer func() {
+			if r := recover(); r != nil {
+				e, ok := r.(stmError)
+				if !ok {
+					// client apply panicked
+					panic(r)
+				}
+				outc <- stmResponse{nil, e.err}
+			}
+		}()
+		var out stmResponse
+		for {
+			s.reset()
+			if out.err = apply(s); out.err != nil {
+				break
+			}
+			if out.resp = s.commit(); out.resp != nil {
+				break
+			}
+		}
+		outc <- out
+	}()
+	r := <-outc
+	return r.resp, r.err
+}
+
+// stm implements repeatable-read software transactional memory over etcd
+type stm struct {
+	client *v3.Client
+	ctx    context.Context
+	// rset holds read key values and revisions
+	rset readSet
+	// wset holds overwritten keys and their values
+	wset writeSet
+	// getOpts are the opts used for gets
+	getOpts []v3.OpOption
+	// conflicts computes the current conflicts on the txn
+	conflicts func() []v3.Cmp
+}
+
+type stmPut struct {
+	val string
+	op  v3.Op
+}
+
+type readSet map[string]*v3.GetResponse
+
+func (rs readSet) add(keys []string, txnresp *v3.TxnResponse) {
+	for i, resp := range txnresp.Responses {
+		rs[keys[i]] = (*v3.GetResponse)(resp.GetResponseRange())
+	}
+}
+
+// first returns the store revision from the first fetch
+func (rs readSet) first() int64 {
+	ret := int64(math.MaxInt64 - 1)
+	for _, resp := range rs {
+		if rev := resp.Header.Revision; rev < ret {
+			ret = rev
+		}
+	}
+	return ret
+}
+
+// cmps guards the txn from updates to read set
+func (rs readSet) cmps() []v3.Cmp {
+	cmps := make([]v3.Cmp, 0, len(rs))
+	for k, rk := range rs {
+		cmps = append(cmps, isKeyCurrent(k, rk))
+	}
+	return cmps
+}
+
+type writeSet map[string]stmPut
+
+func (ws writeSet) get(keys ...string) *stmPut {
+	for _, key := range keys {
+		if wv, ok := ws[key]; ok {
+			return &wv
+		}
+	}
+	return nil
+}
+
+// cmps returns a cmp list testing no writes have happened past rev
+func (ws writeSet) cmps(rev int64) []v3.Cmp {
+	cmps := make([]v3.Cmp, 0, len(ws))
+	for key := range ws {
+		cmps = append(cmps, v3.Compare(v3.ModRevision(key), "<", rev))
+	}
+	return cmps
+}
+
+// puts is the list of ops for all pending writes
+func (ws writeSet) puts() []v3.Op {
+	puts := make([]v3.Op, 0, len(ws))
+	for _, v := range ws {
+		puts = append(puts, v.op)
+	}
+	return puts
+}
+
+func (s *stm) Get(keys ...string) string {
+	if wv := s.wset.get(keys...); wv != nil {
+		return wv.val
+	}
+	return respToValue(s.fetch(keys...))
+}
+
+func (s *stm) Put(key, val string, opts ...v3.OpOption) {
+	s.wset[key] = stmPut{val, v3.OpPut(key, val, opts...)}
+}
+
+func (s *stm) Del(key string) { s.wset[key] = stmPut{"", v3.OpDelete(key)} }
+
+func (s *stm) Rev(key string) int64 {
+	if resp := s.fetch(key); resp != nil && len(resp.Kvs) != 0 {
+		return resp.Kvs[0].ModRevision
+	}
+	return 0
+}
+
+func (s *stm) commit() *v3.TxnResponse {
+	txnresp, err := s.client.Txn(s.ctx).If(s.conflicts()...).Then(s.wset.puts()...).Commit()
+	if err != nil {
+		panic(stmError{err})
+	}
+	if txnresp.Succeeded {
+		return txnresp
+	}
+	return nil
+}
+
+func (s *stm) fetch(keys ...string) *v3.GetResponse {
+	if len(keys) == 0 {
+		return nil
+	}
+	ops := make([]v3.Op, len(keys))
+	for i, key := range keys {
+		if resp, ok := s.rset[key]; ok {
+			return resp
+		}
+		ops[i] = v3.OpGet(key, s.getOpts...)
+	}
+	txnresp, err := s.client.Txn(s.ctx).Then(ops...).Commit()
+	if err != nil {
+		panic(stmError{err})
+	}
+	s.rset.add(keys, txnresp)
+	return (*v3.GetResponse)(txnresp.Responses[0].GetResponseRange())
+}
+
+func (s *stm) reset() {
+	s.rset = make(map[string]*v3.GetResponse)
+	s.wset = make(map[string]stmPut)
+}
+
+type stmSerializable struct {
+	stm
+	prefetch map[string]*v3.GetResponse
+}
+
+func (s *stmSerializable) Get(keys ...string) string {
+	if wv := s.wset.get(keys...); wv != nil {
+		return wv.val
+	}
+	firstRead := len(s.rset) == 0
+	for _, key := range keys {
+		if resp, ok := s.prefetch[key]; ok {
+			delete(s.prefetch, key)
+			s.rset[key] = resp
+		}
+	}
+	resp := s.stm.fetch(keys...)
+	if firstRead {
+		// txn's base revision is defined by the first read
+		s.getOpts = []v3.OpOption{
+			v3.WithRev(resp.Header.Revision),
+			v3.WithSerializable(),
+		}
+	}
+	return respToValue(resp)
+}
+
+func (s *stmSerializable) Rev(key string) int64 {
+	s.Get(key)
+	return s.stm.Rev(key)
+}
+
+func (s *stmSerializable) gets() ([]string, []v3.Op) {
+	keys := make([]string, 0, len(s.rset))
+	ops := make([]v3.Op, 0, len(s.rset))
+	for k := range s.rset {
+		keys = append(keys, k)
+		ops = append(ops, v3.OpGet(k))
+	}
+	return keys, ops
+}
+
+func (s *stmSerializable) commit() *v3.TxnResponse {
+	keys, getops := s.gets()
+	txn := s.client.Txn(s.ctx).If(s.conflicts()...).Then(s.wset.puts()...)
+	// use Else to prefetch keys in case of conflict to save a round trip
+	txnresp, err := txn.Else(getops...).Commit()
+	if err != nil {
+		panic(stmError{err})
+	}
+	if txnresp.Succeeded {
+		return txnresp
+	}
+	// load prefetch with Else data
+	s.rset.add(keys, txnresp)
+	s.prefetch = s.rset
+	s.getOpts = nil
+	return nil
+}
+
+func isKeyCurrent(k string, r *v3.GetResponse) v3.Cmp {
+	if len(r.Kvs) != 0 {
+		return v3.Compare(v3.ModRevision(k), "=", r.Kvs[0].ModRevision)
+	}
+	return v3.Compare(v3.ModRevision(k), "=", 0)
+}
+
+func respToValue(resp *v3.GetResponse) string {
+	if resp == nil || len(resp.Kvs) == 0 {
+		return ""
+	}
+	return string(resp.Kvs[0].Value)
+}
+
+// NewSTMRepeatable is deprecated.
+func NewSTMRepeatable(ctx context.Context, c *v3.Client, apply func(STM) error) (*v3.TxnResponse, error) {
+	return NewSTM(c, apply, WithAbortContext(ctx), WithIsolation(RepeatableReads))
+}
+
+// NewSTMSerializable is deprecated.
+func NewSTMSerializable(ctx context.Context, c *v3.Client, apply func(STM) error) (*v3.TxnResponse, error) {
+	return NewSTM(c, apply, WithAbortContext(ctx), WithIsolation(Serializable))
+}
+
+// NewSTMReadCommitted is deprecated.
+func NewSTMReadCommitted(ctx context.Context, c *v3.Client, apply func(STM) error) (*v3.TxnResponse, error) {
+	return NewSTM(c, apply, WithAbortContext(ctx), WithIsolation(ReadCommitted))
+}

vendor/github.com/coreos/etcd/clientv3/config.go

@@ -0,0 +1,54 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"crypto/tls"
+	"time"
+
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+)
+
+type Config struct {
+	// Endpoints is a list of URLs.
+	Endpoints []string `json:"endpoints"`
+
+	// AutoSyncInterval is the interval to update endpoints with its latest members.
+	// 0 disables auto-sync. By default auto-sync is disabled.
+	AutoSyncInterval time.Duration `json:"auto-sync-interval"`
+
+	// DialTimeout is the timeout for failing to establish a connection.
+	DialTimeout time.Duration `json:"dial-timeout"`
+
+	// TLS holds the client secure credentials, if any.
+	TLS *tls.Config
+
+	// Username is a username for authentication.
+	Username string `json:"username"`
+
+	// Password is a password for authentication.
+	Password string `json:"password"`
+
+	// RejectOldCluster when set will refuse to create a client against an outdated cluster.
+	RejectOldCluster bool `json:"reject-old-cluster"`
+
+	// DialOptions is a list of dial options for the grpc client (e.g., for interceptors).
+	DialOptions []grpc.DialOption
+
+	// Context is the default client context; it can be used to cancel grpc dial out and
+	// other operations that do not have an explicit context.
+	Context context.Context
+}

vendor/github.com/coreos/etcd/clientv3/doc.go

@@ -0,0 +1,64 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Package clientv3 implements the official Go etcd client for v3.
+//
+// Create client using `clientv3.New`:
+//
+//	cli, err := clientv3.New(clientv3.Config{
+//		Endpoints:   []string{"localhost:2379", "localhost:22379", "localhost:32379"},
+//		DialTimeout: 5 * time.Second,
+//	})
+//	if err != nil {
+//		// handle error!
+//	}
+//	defer cli.Close()
+//
+// Make sure to close the client after using it. If the client is not closed, the
+// connection will have leaky goroutines.
+//
+// To specify client request timeout, pass context.WithTimeout to APIs:
+//
+//	ctx, cancel := context.WithTimeout(context.Background(), timeout)
+//	resp, err := kvc.Put(ctx, "sample_key", "sample_value")
+//	cancel()
+//	if err != nil {
+//	    // handle error!
+//	}
+//	// use the response
+//
+// The Client has internal state (watchers and leases), so Clients should be reused instead of created as needed.
+// Clients are safe for concurrent use by multiple goroutines.
+//
+// etcd client returns 2 types of errors:
+//
+//	1. context error: canceled or deadline exceeded.
+//	2. gRPC error: see https://github.com/coreos/etcd/blob/master/etcdserver/api/v3rpc/rpctypes/error.go
+//
+// Here is the example code to handle client errors:
+//
+//	resp, err := kvc.Put(ctx, "", "")
+//	if err != nil {
+//		if err == context.Canceled {
+//			// ctx is canceled by another routine
+//		} else if err == context.DeadlineExceeded {
+//			// ctx is attached with a deadline and it exceeded
+//		} else if verr, ok := err.(*v3rpc.ErrEmptyKey); ok {
+//			// process (verr.Errors)
+//		} else {
+//			// bad cluster endpoints, which are not etcd servers
+//		}
+//	}
+//
+package clientv3

vendor/github.com/coreos/etcd/clientv3/kv.go

@@ -0,0 +1,162 @@
+// Copyright 2015 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+)
+
+type (
+	CompactResponse pb.CompactionResponse
+	PutResponse     pb.PutResponse
+	GetResponse     pb.RangeResponse
+	DeleteResponse  pb.DeleteRangeResponse
+	TxnResponse     pb.TxnResponse
+)
+
+type KV interface {
+	// Put puts a key-value pair into etcd.
+	// Note that key,value can be plain bytes array and string is
+	// an immutable representation of that bytes array.
+	// To get a string of bytes, do string([]byte{0x10, 0x20}).
+	Put(ctx context.Context, key, val string, opts ...OpOption) (*PutResponse, error)
+
+	// Get retrieves keys.
+	// By default, Get will return the value for "key", if any.
+	// When passed WithRange(end), Get will return the keys in the range [key, end).
+	// When passed WithFromKey(), Get returns keys greater than or equal to key.
+	// When passed WithRev(rev) with rev > 0, Get retrieves keys at the given revision;
+	// if the required revision is compacted, the request will fail with ErrCompacted .
+	// When passed WithLimit(limit), the number of returned keys is bounded by limit.
+	// When passed WithSort(), the keys will be sorted.
+	Get(ctx context.Context, key string, opts ...OpOption) (*GetResponse, error)
+
+	// Delete deletes a key, or optionally using WithRange(end), [key, end).
+	Delete(ctx context.Context, key string, opts ...OpOption) (*DeleteResponse, error)
+
+	// Compact compacts etcd KV history before the given rev.
+	Compact(ctx context.Context, rev int64, opts ...CompactOption) (*CompactResponse, error)
+
+	// Do applies a single Op on KV without a transaction.
+	// Do is useful when creating arbitrary operations to be issued at a
+	// later time; the user can range over the operations, calling Do to
+	// execute them. Get/Put/Delete, on the other hand, are best suited
+	// for when the operation should be issued at the time of declaration.
+	Do(ctx context.Context, op Op) (OpResponse, error)
+
+	// Txn creates a transaction.
+	Txn(ctx context.Context) Txn
+}
+
+type OpResponse struct {
+	put *PutResponse
+	get *GetResponse
+	del *DeleteResponse
+}
+
+func (op OpResponse) Put() *PutResponse    { return op.put }
+func (op OpResponse) Get() *GetResponse    { return op.get }
+func (op OpResponse) Del() *DeleteResponse { return op.del }
+
+type kv struct {
+	remote pb.KVClient
+}
+
+func NewKV(c *Client) KV {
+	return &kv{remote: RetryKVClient(c)}
+}
+
+func NewKVFromKVClient(remote pb.KVClient) KV {
+	return &kv{remote: remote}
+}
+
+func (kv *kv) Put(ctx context.Context, key, val string, opts ...OpOption) (*PutResponse, error) {
+	r, err := kv.Do(ctx, OpPut(key, val, opts...))
+	return r.put, toErr(ctx, err)
+}
+
+func (kv *kv) Get(ctx context.Context, key string, opts ...OpOption) (*GetResponse, error) {
+	r, err := kv.Do(ctx, OpGet(key, opts...))
+	return r.get, toErr(ctx, err)
+}
+
+func (kv *kv) Delete(ctx context.Context, key string, opts ...OpOption) (*DeleteResponse, error) {
+	r, err := kv.Do(ctx, OpDelete(key, opts...))
+	return r.del, toErr(ctx, err)
+}
+
+func (kv *kv) Compact(ctx context.Context, rev int64, opts ...CompactOption) (*CompactResponse, error) {
+	resp, err := kv.remote.Compact(ctx, OpCompact(rev, opts...).toRequest())
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	return (*CompactResponse)(resp), err
+}
+
+func (kv *kv) Txn(ctx context.Context) Txn {
+	return &txn{
+		kv:  kv,
+		ctx: ctx,
+	}
+}
+
+func (kv *kv) Do(ctx context.Context, op Op) (OpResponse, error) {
+	for {
+		resp, err := kv.do(ctx, op)
+		if err == nil {
+			return resp, nil
+		}
+
+		if isHaltErr(ctx, err) {
+			return resp, toErr(ctx, err)
+		}
+		// do not retry on modifications
+		if op.isWrite() {
+			return resp, toErr(ctx, err)
+		}
+	}
+}
+
+func (kv *kv) do(ctx context.Context, op Op) (OpResponse, error) {
+	var err error
+	switch op.t {
+	// TODO: handle other ops
+	case tRange:
+		var resp *pb.RangeResponse
+		resp, err = kv.remote.Range(ctx, op.toRangeRequest(), grpc.FailFast(false))
+		if err == nil {
+			return OpResponse{get: (*GetResponse)(resp)}, nil
+		}
+	case tPut:
+		var resp *pb.PutResponse
+		r := &pb.PutRequest{Key: op.key, Value: op.val, Lease: int64(op.leaseID), PrevKv: op.prevKV, IgnoreValue: op.ignoreValue, IgnoreLease: op.ignoreLease}
+		resp, err = kv.remote.Put(ctx, r)
+		if err == nil {
+			return OpResponse{put: (*PutResponse)(resp)}, nil
+		}
+	case tDeleteRange:
+		var resp *pb.DeleteRangeResponse
+		r := &pb.DeleteRangeRequest{Key: op.key, RangeEnd: op.end, PrevKv: op.prevKV}
+		resp, err = kv.remote.DeleteRange(ctx, r)
+		if err == nil {
+			return OpResponse{del: (*DeleteResponse)(resp)}, nil
+		}
+	default:
+		panic("Unknown op")
+	}
+	return OpResponse{}, err
+}

vendor/github.com/coreos/etcd/clientv3/lease.go

@@ -0,0 +1,547 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"sync"
+	"time"
+
+	"github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/metadata"
+)
+
+type (
+	LeaseRevokeResponse pb.LeaseRevokeResponse
+	LeaseID             int64
+)
+
+// LeaseGrantResponse is used to convert the protobuf grant response.
+type LeaseGrantResponse struct {
+	*pb.ResponseHeader
+	ID    LeaseID
+	TTL   int64
+	Error string
+}
+
+// LeaseKeepAliveResponse is used to convert the protobuf keepalive response.
+type LeaseKeepAliveResponse struct {
+	*pb.ResponseHeader
+	ID  LeaseID
+	TTL int64
+}
+
+// LeaseTimeToLiveResponse is used to convert the protobuf lease timetolive response.
+type LeaseTimeToLiveResponse struct {
+	*pb.ResponseHeader
+	ID LeaseID `json:"id"`
+
+	// TTL is the remaining TTL in seconds for the lease; the lease will expire in under TTL+1 seconds.
+	TTL int64 `json:"ttl"`
+
+	// GrantedTTL is the initial granted time in seconds upon lease creation/renewal.
+	GrantedTTL int64 `json:"granted-ttl"`
+
+	// Keys is the list of keys attached to this lease.
+	Keys [][]byte `json:"keys"`
+}
+
+const (
+	// defaultTTL is the assumed lease TTL used for the first keepalive
+	// deadline before the actual TTL is known to the client.
+	defaultTTL = 5 * time.Second
+	// a small buffer to store unsent lease responses.
+	leaseResponseChSize = 16
+	// NoLease is a lease ID for the absence of a lease.
+	NoLease LeaseID = 0
+
+	// retryConnWait is how long to wait before retrying request due to an error
+	retryConnWait = 500 * time.Millisecond
+)
+
+// ErrKeepAliveHalted is returned if client keep alive loop halts with an unexpected error.
+//
+// This usually means that automatic lease renewal via KeepAlive is broken, but KeepAliveOnce will still work as expected.
+type ErrKeepAliveHalted struct {
+	Reason error
+}
+
+func (e ErrKeepAliveHalted) Error() string {
+	s := "etcdclient: leases keep alive halted"
+	if e.Reason != nil {
+		s += ": " + e.Reason.Error()
+	}
+	return s
+}
+
+type Lease interface {
+	// Grant creates a new lease.
+	Grant(ctx context.Context, ttl int64) (*LeaseGrantResponse, error)
+
+	// Revoke revokes the given lease.
+	Revoke(ctx context.Context, id LeaseID) (*LeaseRevokeResponse, error)
+
+	// TimeToLive retrieves the lease information of the given lease ID.
+	TimeToLive(ctx context.Context, id LeaseID, opts ...LeaseOption) (*LeaseTimeToLiveResponse, error)
+
+	// KeepAlive keeps the given lease alive forever.
+	KeepAlive(ctx context.Context, id LeaseID) (<-chan *LeaseKeepAliveResponse, error)
+
+	// KeepAliveOnce renews the lease once. In most of the cases, Keepalive
+	// should be used instead of KeepAliveOnce.
+	KeepAliveOnce(ctx context.Context, id LeaseID) (*LeaseKeepAliveResponse, error)
+
+	// Close releases all resources Lease keeps for efficient communication
+	// with the etcd server.
+	Close() error
+}
+
+type lessor struct {
+	mu sync.Mutex // guards all fields
+
+	// donec is closed and loopErr is set when recvKeepAliveLoop stops
+	donec   chan struct{}
+	loopErr error
+
+	remote pb.LeaseClient
+
+	stream       pb.Lease_LeaseKeepAliveClient
+	streamCancel context.CancelFunc
+
+	stopCtx    context.Context
+	stopCancel context.CancelFunc
+
+	keepAlives map[LeaseID]*keepAlive
+
+	// firstKeepAliveTimeout is the timeout for the first keepalive request
+	// before the actual TTL is known to the lease client
+	firstKeepAliveTimeout time.Duration
+
+	// firstKeepAliveOnce ensures stream starts after first KeepAlive call.
+	firstKeepAliveOnce sync.Once
+}
+
+// keepAlive multiplexes a keepalive for a lease over multiple channels
+type keepAlive struct {
+	chs  []chan<- *LeaseKeepAliveResponse
+	ctxs []context.Context
+	// deadline is the time the keep alive channels close if no response
+	deadline time.Time
+	// nextKeepAlive is when to send the next keep alive message
+	nextKeepAlive time.Time
+	// donec is closed on lease revoke, expiration, or cancel.
+	donec chan struct{}
+}
+
+func NewLease(c *Client) Lease {
+	return NewLeaseFromLeaseClient(RetryLeaseClient(c), c.cfg.DialTimeout+time.Second)
+}
+
+func NewLeaseFromLeaseClient(remote pb.LeaseClient, keepAliveTimeout time.Duration) Lease {
+	l := &lessor{
+		donec:                 make(chan struct{}),
+		keepAlives:            make(map[LeaseID]*keepAlive),
+		remote:                remote,
+		firstKeepAliveTimeout: keepAliveTimeout,
+	}
+	if l.firstKeepAliveTimeout == time.Second {
+		l.firstKeepAliveTimeout = defaultTTL
+	}
+	reqLeaderCtx := WithRequireLeader(context.Background())
+	l.stopCtx, l.stopCancel = context.WithCancel(reqLeaderCtx)
+	return l
+}
+
+func (l *lessor) Grant(ctx context.Context, ttl int64) (*LeaseGrantResponse, error) {
+	for {
+		r := &pb.LeaseGrantRequest{TTL: ttl}
+		resp, err := l.remote.LeaseGrant(ctx, r)
+		if err == nil {
+			gresp := &LeaseGrantResponse{
+				ResponseHeader: resp.GetHeader(),
+				ID:             LeaseID(resp.ID),
+				TTL:            resp.TTL,
+				Error:          resp.Error,
+			}
+			return gresp, nil
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}
+
+func (l *lessor) Revoke(ctx context.Context, id LeaseID) (*LeaseRevokeResponse, error) {
+	for {
+		r := &pb.LeaseRevokeRequest{ID: int64(id)}
+		resp, err := l.remote.LeaseRevoke(ctx, r)
+
+		if err == nil {
+			return (*LeaseRevokeResponse)(resp), nil
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}
+
+func (l *lessor) TimeToLive(ctx context.Context, id LeaseID, opts ...LeaseOption) (*LeaseTimeToLiveResponse, error) {
+	for {
+		r := toLeaseTimeToLiveRequest(id, opts...)
+		resp, err := l.remote.LeaseTimeToLive(ctx, r, grpc.FailFast(false))
+		if err == nil {
+			gresp := &LeaseTimeToLiveResponse{
+				ResponseHeader: resp.GetHeader(),
+				ID:             LeaseID(resp.ID),
+				TTL:            resp.TTL,
+				GrantedTTL:     resp.GrantedTTL,
+				Keys:           resp.Keys,
+			}
+			return gresp, nil
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}
+
+func (l *lessor) KeepAlive(ctx context.Context, id LeaseID) (<-chan *LeaseKeepAliveResponse, error) {
+	ch := make(chan *LeaseKeepAliveResponse, leaseResponseChSize)
+
+	l.mu.Lock()
+	// ensure that recvKeepAliveLoop is still running
+	select {
+	case <-l.donec:
+		err := l.loopErr
+		l.mu.Unlock()
+		close(ch)
+		return ch, ErrKeepAliveHalted{Reason: err}
+	default:
+	}
+	ka, ok := l.keepAlives[id]
+	if !ok {
+		// create fresh keep alive
+		ka = &keepAlive{
+			chs:           []chan<- *LeaseKeepAliveResponse{ch},
+			ctxs:          []context.Context{ctx},
+			deadline:      time.Now().Add(l.firstKeepAliveTimeout),
+			nextKeepAlive: time.Now(),
+			donec:         make(chan struct{}),
+		}
+		l.keepAlives[id] = ka
+	} else {
+		// add channel and context to existing keep alive
+		ka.ctxs = append(ka.ctxs, ctx)
+		ka.chs = append(ka.chs, ch)
+	}
+	l.mu.Unlock()
+
+	go l.keepAliveCtxCloser(id, ctx, ka.donec)
+	l.firstKeepAliveOnce.Do(func() {
+		go l.recvKeepAliveLoop()
+		go l.deadlineLoop()
+	})
+
+	return ch, nil
+}
+
+func (l *lessor) KeepAliveOnce(ctx context.Context, id LeaseID) (*LeaseKeepAliveResponse, error) {
+	for {
+		resp, err := l.keepAliveOnce(ctx, id)
+		if err == nil {
+			if resp.TTL <= 0 {
+				err = rpctypes.ErrLeaseNotFound
+			}
+			return resp, err
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}
+
+func (l *lessor) Close() error {
+	l.stopCancel()
+	// close for synchronous teardown if stream goroutines never launched
+	l.firstKeepAliveOnce.Do(func() { close(l.donec) })
+	<-l.donec
+	return nil
+}
+
+func (l *lessor) keepAliveCtxCloser(id LeaseID, ctx context.Context, donec <-chan struct{}) {
+	select {
+	case <-donec:
+		return
+	case <-l.donec:
+		return
+	case <-ctx.Done():
+	}
+
+	l.mu.Lock()
+	defer l.mu.Unlock()
+
+	ka, ok := l.keepAlives[id]
+	if !ok {
+		return
+	}
+
+	// close channel and remove context if still associated with keep alive
+	for i, c := range ka.ctxs {
+		if c == ctx {
+			close(ka.chs[i])
+			ka.ctxs = append(ka.ctxs[:i], ka.ctxs[i+1:]...)
+			ka.chs = append(ka.chs[:i], ka.chs[i+1:]...)
+			break
+		}
+	}
+	// remove if no one more listeners
+	if len(ka.chs) == 0 {
+		delete(l.keepAlives, id)
+	}
+}
+
+// closeRequireLeader scans all keep alives for ctxs that have require leader
+// and closes the associated channels.
+func (l *lessor) closeRequireLeader() {
+	l.mu.Lock()
+	defer l.mu.Unlock()
+	for _, ka := range l.keepAlives {
+		reqIdxs := 0
+		// find all required leader channels, close, mark as nil
+		for i, ctx := range ka.ctxs {
+			md, ok := metadata.FromContext(ctx)
+			if !ok {
+				continue
+			}
+			ks := md[rpctypes.MetadataRequireLeaderKey]
+			if len(ks) < 1 || ks[0] != rpctypes.MetadataHasLeader {
+				continue
+			}
+			close(ka.chs[i])
+			ka.chs[i] = nil
+			reqIdxs++
+		}
+		if reqIdxs == 0 {
+			continue
+		}
+		// remove all channels that required a leader from keepalive
+		newChs := make([]chan<- *LeaseKeepAliveResponse, len(ka.chs)-reqIdxs)
+		newCtxs := make([]context.Context, len(newChs))
+		newIdx := 0
+		for i := range ka.chs {
+			if ka.chs[i] == nil {
+				continue
+			}
+			newChs[newIdx], newCtxs[newIdx] = ka.chs[i], ka.ctxs[newIdx]
+			newIdx++
+		}
+		ka.chs, ka.ctxs = newChs, newCtxs
+	}
+}
+
+func (l *lessor) keepAliveOnce(ctx context.Context, id LeaseID) (*LeaseKeepAliveResponse, error) {
+	cctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+
+	stream, err := l.remote.LeaseKeepAlive(cctx, grpc.FailFast(false))
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+
+	err = stream.Send(&pb.LeaseKeepAliveRequest{ID: int64(id)})
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+
+	resp, rerr := stream.Recv()
+	if rerr != nil {
+		return nil, toErr(ctx, rerr)
+	}
+
+	karesp := &LeaseKeepAliveResponse{
+		ResponseHeader: resp.GetHeader(),
+		ID:             LeaseID(resp.ID),
+		TTL:            resp.TTL,
+	}
+	return karesp, nil
+}
+
+func (l *lessor) recvKeepAliveLoop() (gerr error) {
+	defer func() {
+		l.mu.Lock()
+		close(l.donec)
+		l.loopErr = gerr
+		for _, ka := range l.keepAlives {
+			ka.Close()
+		}
+		l.keepAlives = make(map[LeaseID]*keepAlive)
+		l.mu.Unlock()
+	}()
+
+	for {
+		stream, err := l.resetRecv()
+		if err != nil {
+			if canceledByCaller(l.stopCtx, err) {
+				return err
+			}
+		} else {
+			for {
+				resp, err := stream.Recv()
+
+				if err != nil {
+					if canceledByCaller(l.stopCtx, err) {
+						return err
+					}
+
+					if toErr(l.stopCtx, err) == rpctypes.ErrNoLeader {
+						l.closeRequireLeader()
+					}
+					break
+				}
+
+				l.recvKeepAlive(resp)
+			}
+		}
+
+		select {
+		case <-time.After(retryConnWait):
+			continue
+		case <-l.stopCtx.Done():
+			return l.stopCtx.Err()
+		}
+	}
+}
+
+// resetRecv opens a new lease stream and starts sending LeaseKeepAliveRequests
+func (l *lessor) resetRecv() (pb.Lease_LeaseKeepAliveClient, error) {
+	sctx, cancel := context.WithCancel(l.stopCtx)
+	stream, err := l.remote.LeaseKeepAlive(sctx, grpc.FailFast(false))
+	if err != nil {
+		cancel()
+		return nil, err
+	}
+
+	l.mu.Lock()
+	defer l.mu.Unlock()
+	if l.stream != nil && l.streamCancel != nil {
+		l.streamCancel()
+	}
+
+	l.streamCancel = cancel
+	l.stream = stream
+
+	go l.sendKeepAliveLoop(stream)
+	return stream, nil
+}
+
+// recvKeepAlive updates a lease based on its LeaseKeepAliveResponse
+func (l *lessor) recvKeepAlive(resp *pb.LeaseKeepAliveResponse) {
+	karesp := &LeaseKeepAliveResponse{
+		ResponseHeader: resp.GetHeader(),
+		ID:             LeaseID(resp.ID),
+		TTL:            resp.TTL,
+	}
+
+	l.mu.Lock()
+	defer l.mu.Unlock()
+
+	ka, ok := l.keepAlives[karesp.ID]
+	if !ok {
+		return
+	}
+
+	if karesp.TTL <= 0 {
+		// lease expired; close all keep alive channels
+		delete(l.keepAlives, karesp.ID)
+		ka.Close()
+		return
+	}
+
+	// send update to all channels
+	nextKeepAlive := time.Now().Add((time.Duration(karesp.TTL) * time.Second) / 3.0)
+	ka.deadline = time.Now().Add(time.Duration(karesp.TTL) * time.Second)
+	for _, ch := range ka.chs {
+		select {
+		case ch <- karesp:
+			ka.nextKeepAlive = nextKeepAlive
+		default:
+		}
+	}
+}
+
+// deadlineLoop reaps any keep alive channels that have not received a response
+// within the lease TTL
+func (l *lessor) deadlineLoop() {
+	for {
+		select {
+		case <-time.After(time.Second):
+		case <-l.donec:
+			return
+		}
+		now := time.Now()
+		l.mu.Lock()
+		for id, ka := range l.keepAlives {
+			if ka.deadline.Before(now) {
+				// waited too long for response; lease may be expired
+				ka.Close()
+				delete(l.keepAlives, id)
+			}
+		}
+		l.mu.Unlock()
+	}
+}
+
+// sendKeepAliveLoop sends LeaseKeepAliveRequests for the lifetime of a lease stream
+func (l *lessor) sendKeepAliveLoop(stream pb.Lease_LeaseKeepAliveClient) {
+	for {
+		var tosend []LeaseID
+
+		now := time.Now()
+		l.mu.Lock()
+		for id, ka := range l.keepAlives {
+			if ka.nextKeepAlive.Before(now) {
+				tosend = append(tosend, id)
+			}
+		}
+		l.mu.Unlock()
+
+		for _, id := range tosend {
+			r := &pb.LeaseKeepAliveRequest{ID: int64(id)}
+			if err := stream.Send(r); err != nil {
+				// TODO do something with this error?
+				return
+			}
+		}
+
+		select {
+		case <-time.After(500 * time.Millisecond):
+		case <-stream.Context().Done():
+			return
+		case <-l.donec:
+			return
+		case <-l.stopCtx.Done():
+			return
+		}
+	}
+}
+
+func (ka *keepAlive) Close() {
+	close(ka.donec)
+	for _, ch := range ka.chs {
+		close(ch)
+	}
+}

vendor/github.com/coreos/etcd/clientv3/logger.go

@@ -0,0 +1,82 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"io/ioutil"
+	"log"
+	"sync"
+
+	"google.golang.org/grpc/grpclog"
+)
+
+// Logger is the logger used by client library.
+// It implements grpclog.Logger interface.
+type Logger grpclog.Logger
+
+var (
+	logger settableLogger
+)
+
+type settableLogger struct {
+	l  grpclog.Logger
+	mu sync.RWMutex
+}
+
+func init() {
+	// disable client side logs by default
+	logger.mu.Lock()
+	logger.l = log.New(ioutil.Discard, "", 0)
+
+	// logger has to override the grpclog at initialization so that
+	// any changes to the grpclog go through logger with locking
+	// instead of through SetLogger
+	//
+	// now updates only happen through settableLogger.set
+	grpclog.SetLogger(&logger)
+	logger.mu.Unlock()
+}
+
+// SetLogger sets client-side Logger. By default, logs are disabled.
+func SetLogger(l Logger) {
+	logger.set(l)
+}
+
+// GetLogger returns the current logger.
+func GetLogger() Logger {
+	return logger.get()
+}
+
+func (s *settableLogger) set(l Logger) {
+	s.mu.Lock()
+	logger.l = l
+	s.mu.Unlock()
+}
+
+func (s *settableLogger) get() Logger {
+	s.mu.RLock()
+	l := logger.l
+	s.mu.RUnlock()
+	return l
+}
+
+// implement the grpclog.Logger interface
+
+func (s *settableLogger) Fatal(args ...interface{})                 { s.get().Fatal(args...) }
+func (s *settableLogger) Fatalf(format string, args ...interface{}) { s.get().Fatalf(format, args...) }
+func (s *settableLogger) Fatalln(args ...interface{})               { s.get().Fatalln(args...) }
+func (s *settableLogger) Print(args ...interface{})                 { s.get().Print(args...) }
+func (s *settableLogger) Printf(format string, args ...interface{}) { s.get().Printf(format, args...) }
+func (s *settableLogger) Println(args ...interface{})               { s.get().Println(args...) }

vendor/github.com/coreos/etcd/clientv3/maintenance.go

@@ -0,0 +1,182 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"io"
+
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+)
+
+type (
+	DefragmentResponse pb.DefragmentResponse
+	AlarmResponse      pb.AlarmResponse
+	AlarmMember        pb.AlarmMember
+	StatusResponse     pb.StatusResponse
+)
+
+type Maintenance interface {
+	// AlarmList gets all active alarms.
+	AlarmList(ctx context.Context) (*AlarmResponse, error)
+
+	// AlarmDisarm disarms a given alarm.
+	AlarmDisarm(ctx context.Context, m *AlarmMember) (*AlarmResponse, error)
+
+	// Defragment defragments storage backend of the etcd member with given endpoint.
+	// Defragment is only needed when deleting a large number of keys and want to reclaim
+	// the resources.
+	// Defragment is an expensive operation. User should avoid defragmenting multiple members
+	// at the same time.
+	// To defragment multiple members in the cluster, user need to call defragment multiple
+	// times with different endpoints.
+	Defragment(ctx context.Context, endpoint string) (*DefragmentResponse, error)
+
+	// Status gets the status of the endpoint.
+	Status(ctx context.Context, endpoint string) (*StatusResponse, error)
+
+	// Snapshot provides a reader for a snapshot of a backend.
+	Snapshot(ctx context.Context) (io.ReadCloser, error)
+}
+
+type maintenance struct {
+	dial   func(endpoint string) (pb.MaintenanceClient, func(), error)
+	remote pb.MaintenanceClient
+}
+
+func NewMaintenance(c *Client) Maintenance {
+	return &maintenance{
+		dial: func(endpoint string) (pb.MaintenanceClient, func(), error) {
+			conn, err := c.dial(endpoint)
+			if err != nil {
+				return nil, nil, err
+			}
+			cancel := func() { conn.Close() }
+			return pb.NewMaintenanceClient(conn), cancel, nil
+		},
+		remote: pb.NewMaintenanceClient(c.conn),
+	}
+}
+
+func NewMaintenanceFromMaintenanceClient(remote pb.MaintenanceClient) Maintenance {
+	return &maintenance{
+		dial: func(string) (pb.MaintenanceClient, func(), error) {
+			return remote, func() {}, nil
+		},
+		remote: remote,
+	}
+}
+
+func (m *maintenance) AlarmList(ctx context.Context) (*AlarmResponse, error) {
+	req := &pb.AlarmRequest{
+		Action:   pb.AlarmRequest_GET,
+		MemberID: 0,                 // all
+		Alarm:    pb.AlarmType_NONE, // all
+	}
+	for {
+		resp, err := m.remote.Alarm(ctx, req, grpc.FailFast(false))
+		if err == nil {
+			return (*AlarmResponse)(resp), nil
+		}
+		if isHaltErr(ctx, err) {
+			return nil, toErr(ctx, err)
+		}
+	}
+}
+
+func (m *maintenance) AlarmDisarm(ctx context.Context, am *AlarmMember) (*AlarmResponse, error) {
+	req := &pb.AlarmRequest{
+		Action:   pb.AlarmRequest_DEACTIVATE,
+		MemberID: am.MemberID,
+		Alarm:    am.Alarm,
+	}
+
+	if req.MemberID == 0 && req.Alarm == pb.AlarmType_NONE {
+		ar, err := m.AlarmList(ctx)
+		if err != nil {
+			return nil, toErr(ctx, err)
+		}
+		ret := AlarmResponse{}
+		for _, am := range ar.Alarms {
+			dresp, derr := m.AlarmDisarm(ctx, (*AlarmMember)(am))
+			if derr != nil {
+				return nil, toErr(ctx, derr)
+			}
+			ret.Alarms = append(ret.Alarms, dresp.Alarms...)
+		}
+		return &ret, nil
+	}
+
+	resp, err := m.remote.Alarm(ctx, req, grpc.FailFast(false))
+	if err == nil {
+		return (*AlarmResponse)(resp), nil
+	}
+	return nil, toErr(ctx, err)
+}
+
+func (m *maintenance) Defragment(ctx context.Context, endpoint string) (*DefragmentResponse, error) {
+	remote, cancel, err := m.dial(endpoint)
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	defer cancel()
+	resp, err := remote.Defragment(ctx, &pb.DefragmentRequest{}, grpc.FailFast(false))
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	return (*DefragmentResponse)(resp), nil
+}
+
+func (m *maintenance) Status(ctx context.Context, endpoint string) (*StatusResponse, error) {
+	remote, cancel, err := m.dial(endpoint)
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	defer cancel()
+	resp, err := remote.Status(ctx, &pb.StatusRequest{}, grpc.FailFast(false))
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+	return (*StatusResponse)(resp), nil
+}
+
+func (m *maintenance) Snapshot(ctx context.Context) (io.ReadCloser, error) {
+	ss, err := m.remote.Snapshot(ctx, &pb.SnapshotRequest{}, grpc.FailFast(false))
+	if err != nil {
+		return nil, toErr(ctx, err)
+	}
+
+	pr, pw := io.Pipe()
+	go func() {
+		for {
+			resp, err := ss.Recv()
+			if err != nil {
+				pw.CloseWithError(err)
+				return
+			}
+			if resp == nil && err == nil {
+				break
+			}
+			if _, werr := pw.Write(resp.Blob); werr != nil {
+				pw.CloseWithError(werr)
+				return
+			}
+		}
+		pw.Close()
+	}()
+	return pr, nil
+}

vendor/github.com/coreos/etcd/clientv3/op.go

@@ -0,0 +1,437 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+
+type opType int
+
+const (
+	// A default Op has opType 0, which is invalid.
+	tRange opType = iota + 1
+	tPut
+	tDeleteRange
+)
+
+var (
+	noPrefixEnd = []byte{0}
+)
+
+// Op represents an Operation that kv can execute.
+type Op struct {
+	t   opType
+	key []byte
+	end []byte
+
+	// for range
+	limit        int64
+	sort         *SortOption
+	serializable bool
+	keysOnly     bool
+	countOnly    bool
+	minModRev    int64
+	maxModRev    int64
+	minCreateRev int64
+	maxCreateRev int64
+
+	// for range, watch
+	rev int64
+
+	// for watch, put, delete
+	prevKV bool
+
+	// for put
+	ignoreValue bool
+	ignoreLease bool
+
+	// progressNotify is for progress updates.
+	progressNotify bool
+	// createdNotify is for created event
+	createdNotify bool
+	// filters for watchers
+	filterPut    bool
+	filterDelete bool
+
+	// for put
+	val     []byte
+	leaseID LeaseID
+}
+
+// accesors / mutators
+
+// KeyBytes returns the byte slice holding the Op's key.
+func (op Op) KeyBytes() []byte { return op.key }
+
+// WithKeyBytes sets the byte slice for the Op's key.
+func (op *Op) WithKeyBytes(key []byte) { op.key = key }
+
+// RangeBytes returns the byte slice holding with the Op's range end, if any.
+func (op Op) RangeBytes() []byte { return op.end }
+
+// WithRangeBytes sets the byte slice for the Op's range end.
+func (op *Op) WithRangeBytes(end []byte) { op.end = end }
+
+// ValueBytes returns the byte slice holding the Op's value, if any.
+func (op Op) ValueBytes() []byte { return op.val }
+
+// WithValueBytes sets the byte slice for the Op's value.
+func (op *Op) WithValueBytes(v []byte) { op.val = v }
+
+func (op Op) toRangeRequest() *pb.RangeRequest {
+	if op.t != tRange {
+		panic("op.t != tRange")
+	}
+	r := &pb.RangeRequest{
+		Key:               op.key,
+		RangeEnd:          op.end,
+		Limit:             op.limit,
+		Revision:          op.rev,
+		Serializable:      op.serializable,
+		KeysOnly:          op.keysOnly,
+		CountOnly:         op.countOnly,
+		MinModRevision:    op.minModRev,
+		MaxModRevision:    op.maxModRev,
+		MinCreateRevision: op.minCreateRev,
+		MaxCreateRevision: op.maxCreateRev,
+	}
+	if op.sort != nil {
+		r.SortOrder = pb.RangeRequest_SortOrder(op.sort.Order)
+		r.SortTarget = pb.RangeRequest_SortTarget(op.sort.Target)
+	}
+	return r
+}
+
+func (op Op) toRequestOp() *pb.RequestOp {
+	switch op.t {
+	case tRange:
+		return &pb.RequestOp{Request: &pb.RequestOp_RequestRange{RequestRange: op.toRangeRequest()}}
+	case tPut:
+		r := &pb.PutRequest{Key: op.key, Value: op.val, Lease: int64(op.leaseID), PrevKv: op.prevKV, IgnoreValue: op.ignoreValue, IgnoreLease: op.ignoreLease}
+		return &pb.RequestOp{Request: &pb.RequestOp_RequestPut{RequestPut: r}}
+	case tDeleteRange:
+		r := &pb.DeleteRangeRequest{Key: op.key, RangeEnd: op.end, PrevKv: op.prevKV}
+		return &pb.RequestOp{Request: &pb.RequestOp_RequestDeleteRange{RequestDeleteRange: r}}
+	default:
+		panic("Unknown Op")
+	}
+}
+
+func (op Op) isWrite() bool {
+	return op.t != tRange
+}
+
+func OpGet(key string, opts ...OpOption) Op {
+	ret := Op{t: tRange, key: []byte(key)}
+	ret.applyOpts(opts)
+	return ret
+}
+
+func OpDelete(key string, opts ...OpOption) Op {
+	ret := Op{t: tDeleteRange, key: []byte(key)}
+	ret.applyOpts(opts)
+	switch {
+	case ret.leaseID != 0:
+		panic("unexpected lease in delete")
+	case ret.limit != 0:
+		panic("unexpected limit in delete")
+	case ret.rev != 0:
+		panic("unexpected revision in delete")
+	case ret.sort != nil:
+		panic("unexpected sort in delete")
+	case ret.serializable:
+		panic("unexpected serializable in delete")
+	case ret.countOnly:
+		panic("unexpected countOnly in delete")
+	case ret.minModRev != 0, ret.maxModRev != 0:
+		panic("unexpected mod revision filter in delete")
+	case ret.minCreateRev != 0, ret.maxCreateRev != 0:
+		panic("unexpected create revision filter in delete")
+	case ret.filterDelete, ret.filterPut:
+		panic("unexpected filter in delete")
+	case ret.createdNotify:
+		panic("unexpected createdNotify in delete")
+	}
+	return ret
+}
+
+func OpPut(key, val string, opts ...OpOption) Op {
+	ret := Op{t: tPut, key: []byte(key), val: []byte(val)}
+	ret.applyOpts(opts)
+	switch {
+	case ret.end != nil:
+		panic("unexpected range in put")
+	case ret.limit != 0:
+		panic("unexpected limit in put")
+	case ret.rev != 0:
+		panic("unexpected revision in put")
+	case ret.sort != nil:
+		panic("unexpected sort in put")
+	case ret.serializable:
+		panic("unexpected serializable in put")
+	case ret.countOnly:
+		panic("unexpected countOnly in put")
+	case ret.minModRev != 0, ret.maxModRev != 0:
+		panic("unexpected mod revision filter in put")
+	case ret.minCreateRev != 0, ret.maxCreateRev != 0:
+		panic("unexpected create revision filter in put")
+	case ret.filterDelete, ret.filterPut:
+		panic("unexpected filter in put")
+	case ret.createdNotify:
+		panic("unexpected createdNotify in put")
+	}
+	return ret
+}
+
+func opWatch(key string, opts ...OpOption) Op {
+	ret := Op{t: tRange, key: []byte(key)}
+	ret.applyOpts(opts)
+	switch {
+	case ret.leaseID != 0:
+		panic("unexpected lease in watch")
+	case ret.limit != 0:
+		panic("unexpected limit in watch")
+	case ret.sort != nil:
+		panic("unexpected sort in watch")
+	case ret.serializable:
+		panic("unexpected serializable in watch")
+	case ret.countOnly:
+		panic("unexpected countOnly in watch")
+	case ret.minModRev != 0, ret.maxModRev != 0:
+		panic("unexpected mod revision filter in watch")
+	case ret.minCreateRev != 0, ret.maxCreateRev != 0:
+		panic("unexpected create revision filter in watch")
+	}
+	return ret
+}
+
+func (op *Op) applyOpts(opts []OpOption) {
+	for _, opt := range opts {
+		opt(op)
+	}
+}
+
+// OpOption configures Operations like Get, Put, Delete.
+type OpOption func(*Op)
+
+// WithLease attaches a lease ID to a key in 'Put' request.
+func WithLease(leaseID LeaseID) OpOption {
+	return func(op *Op) { op.leaseID = leaseID }
+}
+
+// WithLimit limits the number of results to return from 'Get' request.
+// If WithLimit is given a 0 limit, it is treated as no limit.
+func WithLimit(n int64) OpOption { return func(op *Op) { op.limit = n } }
+
+// WithRev specifies the store revision for 'Get' request.
+// Or the start revision of 'Watch' request.
+func WithRev(rev int64) OpOption { return func(op *Op) { op.rev = rev } }
+
+// WithSort specifies the ordering in 'Get' request. It requires
+// 'WithRange' and/or 'WithPrefix' to be specified too.
+// 'target' specifies the target to sort by: key, version, revisions, value.
+// 'order' can be either 'SortNone', 'SortAscend', 'SortDescend'.
+func WithSort(target SortTarget, order SortOrder) OpOption {
+	return func(op *Op) {
+		if target == SortByKey && order == SortAscend {
+			// If order != SortNone, server fetches the entire key-space,
+			// and then applies the sort and limit, if provided.
+			// Since current mvcc.Range implementation returns results
+			// sorted by keys in lexicographically ascending order,
+			// client should ignore SortOrder if the target is SortByKey.
+			order = SortNone
+		}
+		op.sort = &SortOption{target, order}
+	}
+}
+
+// GetPrefixRangeEnd gets the range end of the prefix.
+// 'Get(foo, WithPrefix())' is equal to 'Get(foo, WithRange(GetPrefixRangeEnd(foo))'.
+func GetPrefixRangeEnd(prefix string) string {
+	return string(getPrefix([]byte(prefix)))
+}
+
+func getPrefix(key []byte) []byte {
+	end := make([]byte, len(key))
+	copy(end, key)
+	for i := len(end) - 1; i >= 0; i-- {
+		if end[i] < 0xff {
+			end[i] = end[i] + 1
+			end = end[:i+1]
+			return end
+		}
+	}
+	// next prefix does not exist (e.g., 0xffff);
+	// default to WithFromKey policy
+	return noPrefixEnd
+}
+
+// WithPrefix enables 'Get', 'Delete', or 'Watch' requests to operate
+// on the keys with matching prefix. For example, 'Get(foo, WithPrefix())'
+// can return 'foo1', 'foo2', and so on.
+func WithPrefix() OpOption {
+	return func(op *Op) {
+		if len(op.key) == 0 {
+			op.key, op.end = []byte{0}, []byte{0}
+			return
+		}
+		op.end = getPrefix(op.key)
+	}
+}
+
+// WithRange specifies the range of 'Get', 'Delete', 'Watch' requests.
+// For example, 'Get' requests with 'WithRange(end)' returns
+// the keys in the range [key, end).
+// endKey must be lexicographically greater than start key.
+func WithRange(endKey string) OpOption {
+	return func(op *Op) { op.end = []byte(endKey) }
+}
+
+// WithFromKey specifies the range of 'Get', 'Delete', 'Watch' requests
+// to be equal or greater than the key in the argument.
+func WithFromKey() OpOption { return WithRange("\x00") }
+
+// WithSerializable makes 'Get' request serializable. By default,
+// it's linearizable. Serializable requests are better for lower latency
+// requirement.
+func WithSerializable() OpOption {
+	return func(op *Op) { op.serializable = true }
+}
+
+// WithKeysOnly makes the 'Get' request return only the keys and the corresponding
+// values will be omitted.
+func WithKeysOnly() OpOption {
+	return func(op *Op) { op.keysOnly = true }
+}
+
+// WithCountOnly makes the 'Get' request return only the count of keys.
+func WithCountOnly() OpOption {
+	return func(op *Op) { op.countOnly = true }
+}
+
+// WithMinModRev filters out keys for Get with modification revisions less than the given revision.
+func WithMinModRev(rev int64) OpOption { return func(op *Op) { op.minModRev = rev } }
+
+// WithMaxModRev filters out keys for Get with modification revisions greater than the given revision.
+func WithMaxModRev(rev int64) OpOption { return func(op *Op) { op.maxModRev = rev } }
+
+// WithMinCreateRev filters out keys for Get with creation revisions less than the given revision.
+func WithMinCreateRev(rev int64) OpOption { return func(op *Op) { op.minCreateRev = rev } }
+
+// WithMaxCreateRev filters out keys for Get with creation revisions greater than the given revision.
+func WithMaxCreateRev(rev int64) OpOption { return func(op *Op) { op.maxCreateRev = rev } }
+
+// WithFirstCreate gets the key with the oldest creation revision in the request range.
+func WithFirstCreate() []OpOption { return withTop(SortByCreateRevision, SortAscend) }
+
+// WithLastCreate gets the key with the latest creation revision in the request range.
+func WithLastCreate() []OpOption { return withTop(SortByCreateRevision, SortDescend) }
+
+// WithFirstKey gets the lexically first key in the request range.
+func WithFirstKey() []OpOption { return withTop(SortByKey, SortAscend) }
+
+// WithLastKey gets the lexically last key in the request range.
+func WithLastKey() []OpOption { return withTop(SortByKey, SortDescend) }
+
+// WithFirstRev gets the key with the oldest modification revision in the request range.
+func WithFirstRev() []OpOption { return withTop(SortByModRevision, SortAscend) }
+
+// WithLastRev gets the key with the latest modification revision in the request range.
+func WithLastRev() []OpOption { return withTop(SortByModRevision, SortDescend) }
+
+// withTop gets the first key over the get's prefix given a sort order
+func withTop(target SortTarget, order SortOrder) []OpOption {
+	return []OpOption{WithPrefix(), WithSort(target, order), WithLimit(1)}
+}
+
+// WithProgressNotify makes watch server send periodic progress updates
+// every 10 minutes when there is no incoming events.
+// Progress updates have zero events in WatchResponse.
+func WithProgressNotify() OpOption {
+	return func(op *Op) {
+		op.progressNotify = true
+	}
+}
+
+// WithCreatedNotify makes watch server sends the created event.
+func WithCreatedNotify() OpOption {
+	return func(op *Op) {
+		op.createdNotify = true
+	}
+}
+
+// WithFilterPut discards PUT events from the watcher.
+func WithFilterPut() OpOption {
+	return func(op *Op) { op.filterPut = true }
+}
+
+// WithFilterDelete discards DELETE events from the watcher.
+func WithFilterDelete() OpOption {
+	return func(op *Op) { op.filterDelete = true }
+}
+
+// WithPrevKV gets the previous key-value pair before the event happens. If the previous KV is already compacted,
+// nothing will be returned.
+func WithPrevKV() OpOption {
+	return func(op *Op) {
+		op.prevKV = true
+	}
+}
+
+// WithIgnoreValue updates the key using its current value.
+// Empty value should be passed when ignore_value is set.
+// Returns an error if the key does not exist.
+func WithIgnoreValue() OpOption {
+	return func(op *Op) {
+		op.ignoreValue = true
+	}
+}
+
+// WithIgnoreLease updates the key using its current lease.
+// Empty lease should be passed when ignore_lease is set.
+// Returns an error if the key does not exist.
+func WithIgnoreLease() OpOption {
+	return func(op *Op) {
+		op.ignoreLease = true
+	}
+}
+
+// LeaseOp represents an Operation that lease can execute.
+type LeaseOp struct {
+	id LeaseID
+
+	// for TimeToLive
+	attachedKeys bool
+}
+
+// LeaseOption configures lease operations.
+type LeaseOption func(*LeaseOp)
+
+func (op *LeaseOp) applyOpts(opts []LeaseOption) {
+	for _, opt := range opts {
+		opt(op)
+	}
+}
+
+// WithAttachedKeys requests lease timetolive API to return
+// attached keys of given lease ID.
+func WithAttachedKeys() LeaseOption {
+	return func(op *LeaseOp) { op.attachedKeys = true }
+}
+
+func toLeaseTimeToLiveRequest(id LeaseID, opts ...LeaseOption) *pb.LeaseTimeToLiveRequest {
+	ret := &LeaseOp{id: id}
+	ret.applyOpts(opts)
+	return &pb.LeaseTimeToLiveRequest{ID: int64(id), Keys: ret.attachedKeys}
+}

vendor/github.com/coreos/etcd/clientv3/retry.go

@@ -0,0 +1,293 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+)
+
+type rpcFunc func(ctx context.Context) error
+type retryRpcFunc func(context.Context, rpcFunc) error
+
+func (c *Client) newRetryWrapper() retryRpcFunc {
+	return func(rpcCtx context.Context, f rpcFunc) error {
+		for {
+			err := f(rpcCtx)
+			if err == nil {
+				return nil
+			}
+
+			eErr := rpctypes.Error(err)
+			// always stop retry on etcd errors
+			if _, ok := eErr.(rpctypes.EtcdError); ok {
+				return err
+			}
+
+			// only retry if unavailable
+			if grpc.Code(err) != codes.Unavailable {
+				return err
+			}
+
+			select {
+			case <-c.balancer.ConnectNotify():
+			case <-rpcCtx.Done():
+				return rpcCtx.Err()
+			case <-c.ctx.Done():
+				return c.ctx.Err()
+			}
+		}
+	}
+}
+
+func (c *Client) newAuthRetryWrapper() retryRpcFunc {
+	return func(rpcCtx context.Context, f rpcFunc) error {
+		for {
+			err := f(rpcCtx)
+			if err == nil {
+				return nil
+			}
+
+			// always stop retry on etcd errors other than invalid auth token
+			if rpctypes.Error(err) == rpctypes.ErrInvalidAuthToken {
+				gterr := c.getToken(rpcCtx)
+				if gterr != nil {
+					return err // return the original error for simplicity
+				}
+				continue
+			}
+
+			return err
+		}
+	}
+}
+
+// RetryKVClient implements a KVClient that uses the client's FailFast retry policy.
+func RetryKVClient(c *Client) pb.KVClient {
+	retryWrite := &retryWriteKVClient{pb.NewKVClient(c.conn), c.retryWrapper}
+	return &retryKVClient{&retryWriteKVClient{retryWrite, c.retryAuthWrapper}}
+}
+
+type retryKVClient struct {
+	*retryWriteKVClient
+}
+
+func (rkv *retryKVClient) Range(ctx context.Context, in *pb.RangeRequest, opts ...grpc.CallOption) (resp *pb.RangeResponse, err error) {
+	err = rkv.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rkv.retryWriteKVClient.Range(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+type retryWriteKVClient struct {
+	pb.KVClient
+	retryf retryRpcFunc
+}
+
+func (rkv *retryWriteKVClient) Put(ctx context.Context, in *pb.PutRequest, opts ...grpc.CallOption) (resp *pb.PutResponse, err error) {
+	err = rkv.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rkv.KVClient.Put(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rkv *retryWriteKVClient) DeleteRange(ctx context.Context, in *pb.DeleteRangeRequest, opts ...grpc.CallOption) (resp *pb.DeleteRangeResponse, err error) {
+	err = rkv.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rkv.KVClient.DeleteRange(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rkv *retryWriteKVClient) Txn(ctx context.Context, in *pb.TxnRequest, opts ...grpc.CallOption) (resp *pb.TxnResponse, err error) {
+	err = rkv.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rkv.KVClient.Txn(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rkv *retryWriteKVClient) Compact(ctx context.Context, in *pb.CompactionRequest, opts ...grpc.CallOption) (resp *pb.CompactionResponse, err error) {
+	err = rkv.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rkv.KVClient.Compact(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+type retryLeaseClient struct {
+	pb.LeaseClient
+	retryf retryRpcFunc
+}
+
+// RetryLeaseClient implements a LeaseClient that uses the client's FailFast retry policy.
+func RetryLeaseClient(c *Client) pb.LeaseClient {
+	retry := &retryLeaseClient{pb.NewLeaseClient(c.conn), c.retryWrapper}
+	return &retryLeaseClient{retry, c.retryAuthWrapper}
+}
+
+func (rlc *retryLeaseClient) LeaseGrant(ctx context.Context, in *pb.LeaseGrantRequest, opts ...grpc.CallOption) (resp *pb.LeaseGrantResponse, err error) {
+	err = rlc.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rlc.LeaseClient.LeaseGrant(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+
+}
+
+func (rlc *retryLeaseClient) LeaseRevoke(ctx context.Context, in *pb.LeaseRevokeRequest, opts ...grpc.CallOption) (resp *pb.LeaseRevokeResponse, err error) {
+	err = rlc.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rlc.LeaseClient.LeaseRevoke(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+type retryClusterClient struct {
+	pb.ClusterClient
+	retryf retryRpcFunc
+}
+
+// RetryClusterClient implements a ClusterClient that uses the client's FailFast retry policy.
+func RetryClusterClient(c *Client) pb.ClusterClient {
+	return &retryClusterClient{pb.NewClusterClient(c.conn), c.retryWrapper}
+}
+
+func (rcc *retryClusterClient) MemberAdd(ctx context.Context, in *pb.MemberAddRequest, opts ...grpc.CallOption) (resp *pb.MemberAddResponse, err error) {
+	err = rcc.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rcc.ClusterClient.MemberAdd(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rcc *retryClusterClient) MemberRemove(ctx context.Context, in *pb.MemberRemoveRequest, opts ...grpc.CallOption) (resp *pb.MemberRemoveResponse, err error) {
+	err = rcc.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rcc.ClusterClient.MemberRemove(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rcc *retryClusterClient) MemberUpdate(ctx context.Context, in *pb.MemberUpdateRequest, opts ...grpc.CallOption) (resp *pb.MemberUpdateResponse, err error) {
+	err = rcc.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rcc.ClusterClient.MemberUpdate(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+type retryAuthClient struct {
+	pb.AuthClient
+	retryf retryRpcFunc
+}
+
+// RetryAuthClient implements a AuthClient that uses the client's FailFast retry policy.
+func RetryAuthClient(c *Client) pb.AuthClient {
+	return &retryAuthClient{pb.NewAuthClient(c.conn), c.retryWrapper}
+}
+
+func (rac *retryAuthClient) AuthEnable(ctx context.Context, in *pb.AuthEnableRequest, opts ...grpc.CallOption) (resp *pb.AuthEnableResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.AuthEnable(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) AuthDisable(ctx context.Context, in *pb.AuthDisableRequest, opts ...grpc.CallOption) (resp *pb.AuthDisableResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.AuthDisable(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) UserAdd(ctx context.Context, in *pb.AuthUserAddRequest, opts ...grpc.CallOption) (resp *pb.AuthUserAddResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.UserAdd(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) UserDelete(ctx context.Context, in *pb.AuthUserDeleteRequest, opts ...grpc.CallOption) (resp *pb.AuthUserDeleteResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.UserDelete(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) UserChangePassword(ctx context.Context, in *pb.AuthUserChangePasswordRequest, opts ...grpc.CallOption) (resp *pb.AuthUserChangePasswordResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.UserChangePassword(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) UserGrantRole(ctx context.Context, in *pb.AuthUserGrantRoleRequest, opts ...grpc.CallOption) (resp *pb.AuthUserGrantRoleResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.UserGrantRole(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) UserRevokeRole(ctx context.Context, in *pb.AuthUserRevokeRoleRequest, opts ...grpc.CallOption) (resp *pb.AuthUserRevokeRoleResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.UserRevokeRole(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) RoleAdd(ctx context.Context, in *pb.AuthRoleAddRequest, opts ...grpc.CallOption) (resp *pb.AuthRoleAddResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.RoleAdd(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) RoleDelete(ctx context.Context, in *pb.AuthRoleDeleteRequest, opts ...grpc.CallOption) (resp *pb.AuthRoleDeleteResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.RoleDelete(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) RoleGrantPermission(ctx context.Context, in *pb.AuthRoleGrantPermissionRequest, opts ...grpc.CallOption) (resp *pb.AuthRoleGrantPermissionResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.RoleGrantPermission(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}
+
+func (rac *retryAuthClient) RoleRevokePermission(ctx context.Context, in *pb.AuthRoleRevokePermissionRequest, opts ...grpc.CallOption) (resp *pb.AuthRoleRevokePermissionResponse, err error) {
+	err = rac.retryf(ctx, func(rctx context.Context) error {
+		resp, err = rac.AuthClient.RoleRevokePermission(rctx, in, opts...)
+		return err
+	})
+	return resp, err
+}

vendor/github.com/coreos/etcd/clientv3/sort.go

@@ -0,0 +1,37 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+type SortTarget int
+type SortOrder int
+
+const (
+	SortNone SortOrder = iota
+	SortAscend
+	SortDescend
+)
+
+const (
+	SortByKey SortTarget = iota
+	SortByVersion
+	SortByCreateRevision
+	SortByModRevision
+	SortByValue
+)
+
+type SortOption struct {
+	Target SortTarget
+	Order  SortOrder
+}

vendor/github.com/coreos/etcd/clientv3/txn.go

@@ -0,0 +1,164 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"sync"
+
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+)
+
+// Txn is the interface that wraps mini-transactions.
+//
+//	 Tx.If(
+//	  Compare(Value(k1), ">", v1),
+//	  Compare(Version(k1), "=", 2)
+//	 ).Then(
+//	  OpPut(k2,v2), OpPut(k3,v3)
+//	 ).Else(
+//	  OpPut(k4,v4), OpPut(k5,v5)
+//	 ).Commit()
+//
+type Txn interface {
+	// If takes a list of comparison. If all comparisons passed in succeed,
+	// the operations passed into Then() will be executed. Or the operations
+	// passed into Else() will be executed.
+	If(cs ...Cmp) Txn
+
+	// Then takes a list of operations. The Ops list will be executed, if the
+	// comparisons passed in If() succeed.
+	Then(ops ...Op) Txn
+
+	// Else takes a list of operations. The Ops list will be executed, if the
+	// comparisons passed in If() fail.
+	Else(ops ...Op) Txn
+
+	// Commit tries to commit the transaction.
+	Commit() (*TxnResponse, error)
+}
+
+type txn struct {
+	kv  *kv
+	ctx context.Context
+
+	mu    sync.Mutex
+	cif   bool
+	cthen bool
+	celse bool
+
+	isWrite bool
+
+	cmps []*pb.Compare
+
+	sus []*pb.RequestOp
+	fas []*pb.RequestOp
+}
+
+func (txn *txn) If(cs ...Cmp) Txn {
+	txn.mu.Lock()
+	defer txn.mu.Unlock()
+
+	if txn.cif {
+		panic("cannot call If twice!")
+	}
+
+	if txn.cthen {
+		panic("cannot call If after Then!")
+	}
+
+	if txn.celse {
+		panic("cannot call If after Else!")
+	}
+
+	txn.cif = true
+
+	for i := range cs {
+		txn.cmps = append(txn.cmps, (*pb.Compare)(&cs[i]))
+	}
+
+	return txn
+}
+
+func (txn *txn) Then(ops ...Op) Txn {
+	txn.mu.Lock()
+	defer txn.mu.Unlock()
+
+	if txn.cthen {
+		panic("cannot call Then twice!")
+	}
+	if txn.celse {
+		panic("cannot call Then after Else!")
+	}
+
+	txn.cthen = true
+
+	for _, op := range ops {
+		txn.isWrite = txn.isWrite || op.isWrite()
+		txn.sus = append(txn.sus, op.toRequestOp())
+	}
+
+	return txn
+}
+
+func (txn *txn) Else(ops ...Op) Txn {
+	txn.mu.Lock()
+	defer txn.mu.Unlock()
+
+	if txn.celse {
+		panic("cannot call Else twice!")
+	}
+
+	txn.celse = true
+
+	for _, op := range ops {
+		txn.isWrite = txn.isWrite || op.isWrite()
+		txn.fas = append(txn.fas, op.toRequestOp())
+	}
+
+	return txn
+}
+
+func (txn *txn) Commit() (*TxnResponse, error) {
+	txn.mu.Lock()
+	defer txn.mu.Unlock()
+	for {
+		resp, err := txn.commit()
+		if err == nil {
+			return resp, err
+		}
+		if isHaltErr(txn.ctx, err) {
+			return nil, toErr(txn.ctx, err)
+		}
+		if txn.isWrite {
+			return nil, toErr(txn.ctx, err)
+		}
+	}
+}
+
+func (txn *txn) commit() (*TxnResponse, error) {
+	r := &pb.TxnRequest{Compare: txn.cmps, Success: txn.sus, Failure: txn.fas}
+
+	var opts []grpc.CallOption
+	if !txn.isWrite {
+		opts = []grpc.CallOption{grpc.FailFast(false)}
+	}
+	resp, err := txn.kv.remote.Txn(txn.ctx, r, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return (*TxnResponse)(resp), nil
+}

vendor/github.com/coreos/etcd/clientv3/watch.go

@@ -0,0 +1,797 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package clientv3
+
+import (
+	"fmt"
+	"sync"
+	"time"
+
+	v3rpc "github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes"
+	pb "github.com/coreos/etcd/etcdserver/etcdserverpb"
+	mvccpb "github.com/coreos/etcd/mvcc/mvccpb"
+	"golang.org/x/net/context"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+)
+
+const (
+	EventTypeDelete = mvccpb.DELETE
+	EventTypePut    = mvccpb.PUT
+
+	closeSendErrTimeout = 250 * time.Millisecond
+)
+
+type Event mvccpb.Event
+
+type WatchChan <-chan WatchResponse
+
+type Watcher interface {
+	// Watch watches on a key or prefix. The watched events will be returned
+	// through the returned channel.
+	// If the watch is slow or the required rev is compacted, the watch request
+	// might be canceled from the server-side and the chan will be closed.
+	// 'opts' can be: 'WithRev' and/or 'WithPrefix'.
+	Watch(ctx context.Context, key string, opts ...OpOption) WatchChan
+
+	// Close closes the watcher and cancels all watch requests.
+	Close() error
+}
+
+type WatchResponse struct {
+	Header pb.ResponseHeader
+	Events []*Event
+
+	// CompactRevision is the minimum revision the watcher may receive.
+	CompactRevision int64
+
+	// Canceled is used to indicate watch failure.
+	// If the watch failed and the stream was about to close, before the channel is closed,
+	// the channel sends a final response that has Canceled set to true with a non-nil Err().
+	Canceled bool
+
+	// Created is used to indicate the creation of the watcher.
+	Created bool
+
+	closeErr error
+
+	// cancelReason is a reason of canceling watch
+	cancelReason string
+}
+
+// IsCreate returns true if the event tells that the key is newly created.
+func (e *Event) IsCreate() bool {
+	return e.Type == EventTypePut && e.Kv.CreateRevision == e.Kv.ModRevision
+}
+
+// IsModify returns true if the event tells that a new value is put on existing key.
+func (e *Event) IsModify() bool {
+	return e.Type == EventTypePut && e.Kv.CreateRevision != e.Kv.ModRevision
+}
+
+// Err is the error value if this WatchResponse holds an error.
+func (wr *WatchResponse) Err() error {
+	switch {
+	case wr.closeErr != nil:
+		return v3rpc.Error(wr.closeErr)
+	case wr.CompactRevision != 0:
+		return v3rpc.ErrCompacted
+	case wr.Canceled:
+		if len(wr.cancelReason) != 0 {
+			return v3rpc.Error(grpc.Errorf(codes.FailedPrecondition, "%s", wr.cancelReason))
+		}
+		return v3rpc.ErrFutureRev
+	}
+	return nil
+}
+
+// IsProgressNotify returns true if the WatchResponse is progress notification.
+func (wr *WatchResponse) IsProgressNotify() bool {
+	return len(wr.Events) == 0 && !wr.Canceled && !wr.Created && wr.CompactRevision == 0 && wr.Header.Revision != 0
+}
+
+// watcher implements the Watcher interface
+type watcher struct {
+	remote pb.WatchClient
+
+	// mu protects the grpc streams map
+	mu sync.RWMutex
+
+	// streams holds all the active grpc streams keyed by ctx value.
+	streams map[string]*watchGrpcStream
+}
+
+// watchGrpcStream tracks all watch resources attached to a single grpc stream.
+type watchGrpcStream struct {
+	owner  *watcher
+	remote pb.WatchClient
+
+	// ctx controls internal remote.Watch requests
+	ctx context.Context
+	// ctxKey is the key used when looking up this stream's context
+	ctxKey string
+	cancel context.CancelFunc
+
+	// substreams holds all active watchers on this grpc stream
+	substreams map[int64]*watcherStream
+	// resuming holds all resuming watchers on this grpc stream
+	resuming []*watcherStream
+
+	// reqc sends a watch request from Watch() to the main goroutine
+	reqc chan *watchRequest
+	// respc receives data from the watch client
+	respc chan *pb.WatchResponse
+	// donec closes to broadcast shutdown
+	donec chan struct{}
+	// errc transmits errors from grpc Recv to the watch stream reconn logic
+	errc chan error
+	// closingc gets the watcherStream of closing watchers
+	closingc chan *watcherStream
+	// wg is Done when all substream goroutines have exited
+	wg sync.WaitGroup
+
+	// resumec closes to signal that all substreams should begin resuming
+	resumec chan struct{}
+	// closeErr is the error that closed the watch stream
+	closeErr error
+}
+
+// watchRequest is issued by the subscriber to start a new watcher
+type watchRequest struct {
+	ctx context.Context
+	key string
+	end string
+	rev int64
+	// send created notification event if this field is true
+	createdNotify bool
+	// progressNotify is for progress updates
+	progressNotify bool
+	// filters is the list of events to filter out
+	filters []pb.WatchCreateRequest_FilterType
+	// get the previous key-value pair before the event happens
+	prevKV bool
+	// retc receives a chan WatchResponse once the watcher is established
+	retc chan chan WatchResponse
+}
+
+// watcherStream represents a registered watcher
+type watcherStream struct {
+	// initReq is the request that initiated this request
+	initReq watchRequest
+
+	// outc publishes watch responses to subscriber
+	outc chan WatchResponse
+	// recvc buffers watch responses before publishing
+	recvc chan *WatchResponse
+	// donec closes when the watcherStream goroutine stops.
+	donec chan struct{}
+	// closing is set to true when stream should be scheduled to shutdown.
+	closing bool
+	// id is the registered watch id on the grpc stream
+	id int64
+
+	// buf holds all events received from etcd but not yet consumed by the client
+	buf []*WatchResponse
+}
+
+func NewWatcher(c *Client) Watcher {
+	return NewWatchFromWatchClient(pb.NewWatchClient(c.conn))
+}
+
+func NewWatchFromWatchClient(wc pb.WatchClient) Watcher {
+	return &watcher{
+		remote:  wc,
+		streams: make(map[string]*watchGrpcStream),
+	}
+}
+
+// never closes
+var valCtxCh = make(chan struct{})
+var zeroTime = time.Unix(0, 0)
+
+// ctx with only the values; never Done
+type valCtx struct{ context.Context }
+
+func (vc *valCtx) Deadline() (time.Time, bool) { return zeroTime, false }
+func (vc *valCtx) Done() <-chan struct{}       { return valCtxCh }
+func (vc *valCtx) Err() error                  { return nil }
+
+func (w *watcher) newWatcherGrpcStream(inctx context.Context) *watchGrpcStream {
+	ctx, cancel := context.WithCancel(&valCtx{inctx})
+	wgs := &watchGrpcStream{
+		owner:      w,
+		remote:     w.remote,
+		ctx:        ctx,
+		ctxKey:     fmt.Sprintf("%v", inctx),
+		cancel:     cancel,
+		substreams: make(map[int64]*watcherStream),
+
+		respc:    make(chan *pb.WatchResponse),
+		reqc:     make(chan *watchRequest),
+		donec:    make(chan struct{}),
+		errc:     make(chan error, 1),
+		closingc: make(chan *watcherStream),
+		resumec:  make(chan struct{}),
+	}
+	go wgs.run()
+	return wgs
+}
+
+// Watch posts a watch request to run() and waits for a new watcher channel
+func (w *watcher) Watch(ctx context.Context, key string, opts ...OpOption) WatchChan {
+	ow := opWatch(key, opts...)
+
+	var filters []pb.WatchCreateRequest_FilterType
+	if ow.filterPut {
+		filters = append(filters, pb.WatchCreateRequest_NOPUT)
+	}
+	if ow.filterDelete {
+		filters = append(filters, pb.WatchCreateRequest_NODELETE)
+	}
+
+	wr := &watchRequest{
+		ctx:            ctx,
+		createdNotify:  ow.createdNotify,
+		key:            string(ow.key),
+		end:            string(ow.end),
+		rev:            ow.rev,
+		progressNotify: ow.progressNotify,
+		filters:        filters,
+		prevKV:         ow.prevKV,
+		retc:           make(chan chan WatchResponse, 1),
+	}
+
+	ok := false
+	ctxKey := fmt.Sprintf("%v", ctx)
+
+	// find or allocate appropriate grpc watch stream
+	w.mu.Lock()
+	if w.streams == nil {
+		// closed
+		w.mu.Unlock()
+		ch := make(chan WatchResponse)
+		close(ch)
+		return ch
+	}
+	wgs := w.streams[ctxKey]
+	if wgs == nil {
+		wgs = w.newWatcherGrpcStream(ctx)
+		w.streams[ctxKey] = wgs
+	}
+	donec := wgs.donec
+	reqc := wgs.reqc
+	w.mu.Unlock()
+
+	// couldn't create channel; return closed channel
+	closeCh := make(chan WatchResponse, 1)
+
+	// submit request
+	select {
+	case reqc <- wr:
+		ok = true
+	case <-wr.ctx.Done():
+	case <-donec:
+		if wgs.closeErr != nil {
+			closeCh <- WatchResponse{closeErr: wgs.closeErr}
+			break
+		}
+		// retry; may have dropped stream from no ctxs
+		return w.Watch(ctx, key, opts...)
+	}
+
+	// receive channel
+	if ok {
+		select {
+		case ret := <-wr.retc:
+			return ret
+		case <-ctx.Done():
+		case <-donec:
+			if wgs.closeErr != nil {
+				closeCh <- WatchResponse{closeErr: wgs.closeErr}
+				break
+			}
+			// retry; may have dropped stream from no ctxs
+			return w.Watch(ctx, key, opts...)
+		}
+	}
+
+	close(closeCh)
+	return closeCh
+}
+
+func (w *watcher) Close() (err error) {
+	w.mu.Lock()
+	streams := w.streams
+	w.streams = nil
+	w.mu.Unlock()
+	for _, wgs := range streams {
+		if werr := wgs.Close(); werr != nil {
+			err = werr
+		}
+	}
+	return err
+}
+
+func (w *watchGrpcStream) Close() (err error) {
+	w.cancel()
+	<-w.donec
+	select {
+	case err = <-w.errc:
+	default:
+	}
+	return toErr(w.ctx, err)
+}
+
+func (w *watcher) closeStream(wgs *watchGrpcStream) {
+	w.mu.Lock()
+	close(wgs.donec)
+	wgs.cancel()
+	if w.streams != nil {
+		delete(w.streams, wgs.ctxKey)
+	}
+	w.mu.Unlock()
+}
+
+func (w *watchGrpcStream) addSubstream(resp *pb.WatchResponse, ws *watcherStream) {
+	if resp.WatchId == -1 {
+		// failed; no channel
+		close(ws.recvc)
+		return
+	}
+	ws.id = resp.WatchId
+	w.substreams[ws.id] = ws
+}
+
+func (w *watchGrpcStream) sendCloseSubstream(ws *watcherStream, resp *WatchResponse) {
+	select {
+	case ws.outc <- *resp:
+	case <-ws.initReq.ctx.Done():
+	case <-time.After(closeSendErrTimeout):
+	}
+	close(ws.outc)
+}
+
+func (w *watchGrpcStream) closeSubstream(ws *watcherStream) {
+	// send channel response in case stream was never established
+	select {
+	case ws.initReq.retc <- ws.outc:
+	default:
+	}
+	// close subscriber's channel
+	if closeErr := w.closeErr; closeErr != nil && ws.initReq.ctx.Err() == nil {
+		go w.sendCloseSubstream(ws, &WatchResponse{closeErr: w.closeErr})
+	} else if ws.outc != nil {
+		close(ws.outc)
+	}
+	if ws.id != -1 {
+		delete(w.substreams, ws.id)
+		return
+	}
+	for i := range w.resuming {
+		if w.resuming[i] == ws {
+			w.resuming[i] = nil
+			return
+		}
+	}
+}
+
+// run is the root of the goroutines for managing a watcher client
+func (w *watchGrpcStream) run() {
+	var wc pb.Watch_WatchClient
+	var closeErr error
+
+	// substreams marked to close but goroutine still running; needed for
+	// avoiding double-closing recvc on grpc stream teardown
+	closing := make(map[*watcherStream]struct{})
+
+	defer func() {
+		w.closeErr = closeErr
+		// shutdown substreams and resuming substreams
+		for _, ws := range w.substreams {
+			if _, ok := closing[ws]; !ok {
+				close(ws.recvc)
+				closing[ws] = struct{}{}
+			}
+		}
+		for _, ws := range w.resuming {
+			if _, ok := closing[ws]; ws != nil && !ok {
+				close(ws.recvc)
+				closing[ws] = struct{}{}
+			}
+		}
+		w.joinSubstreams()
+		for range closing {
+			w.closeSubstream(<-w.closingc)
+		}
+		w.wg.Wait()
+		w.owner.closeStream(w)
+	}()
+
+	// start a stream with the etcd grpc server
+	if wc, closeErr = w.newWatchClient(); closeErr != nil {
+		return
+	}
+
+	cancelSet := make(map[int64]struct{})
+
+	for {
+		select {
+		// Watch() requested
+		case wreq := <-w.reqc:
+			outc := make(chan WatchResponse, 1)
+			ws := &watcherStream{
+				initReq: *wreq,
+				id:      -1,
+				outc:    outc,
+				// unbufffered so resumes won't cause repeat events
+				recvc: make(chan *WatchResponse),
+			}
+
+			ws.donec = make(chan struct{})
+			w.wg.Add(1)
+			go w.serveSubstream(ws, w.resumec)
+
+			// queue up for watcher creation/resume
+			w.resuming = append(w.resuming, ws)
+			if len(w.resuming) == 1 {
+				// head of resume queue, can register a new watcher
+				wc.Send(ws.initReq.toPB())
+			}
+		// New events from the watch client
+		case pbresp := <-w.respc:
+			switch {
+			case pbresp.Created:
+				// response to head of queue creation
+				if ws := w.resuming[0]; ws != nil {
+					w.addSubstream(pbresp, ws)
+					w.dispatchEvent(pbresp)
+					w.resuming[0] = nil
+				}
+				if ws := w.nextResume(); ws != nil {
+					wc.Send(ws.initReq.toPB())
+				}
+			case pbresp.Canceled:
+				delete(cancelSet, pbresp.WatchId)
+				if ws, ok := w.substreams[pbresp.WatchId]; ok {
+					// signal to stream goroutine to update closingc
+					close(ws.recvc)
+					closing[ws] = struct{}{}
+				}
+			default:
+				// dispatch to appropriate watch stream
+				if ok := w.dispatchEvent(pbresp); ok {
+					break
+				}
+				// watch response on unexpected watch id; cancel id
+				if _, ok := cancelSet[pbresp.WatchId]; ok {
+					break
+				}
+				cancelSet[pbresp.WatchId] = struct{}{}
+				cr := &pb.WatchRequest_CancelRequest{
+					CancelRequest: &pb.WatchCancelRequest{
+						WatchId: pbresp.WatchId,
+					},
+				}
+				req := &pb.WatchRequest{RequestUnion: cr}
+				wc.Send(req)
+			}
+		// watch client failed to recv; spawn another if possible
+		case err := <-w.errc:
+			if isHaltErr(w.ctx, err) || toErr(w.ctx, err) == v3rpc.ErrNoLeader {
+				closeErr = err
+				return
+			}
+			if wc, closeErr = w.newWatchClient(); closeErr != nil {
+				return
+			}
+			if ws := w.nextResume(); ws != nil {
+				wc.Send(ws.initReq.toPB())
+			}
+			cancelSet = make(map[int64]struct{})
+		case <-w.ctx.Done():
+			return
+		case ws := <-w.closingc:
+			w.closeSubstream(ws)
+			delete(closing, ws)
+			if len(w.substreams)+len(w.resuming) == 0 {
+				// no more watchers on this stream, shutdown
+				return
+			}
+		}
+	}
+}
+
+// nextResume chooses the next resuming to register with the grpc stream. Abandoned
+// streams are marked as nil in the queue since the head must wait for its inflight registration.
+func (w *watchGrpcStream) nextResume() *watcherStream {
+	for len(w.resuming) != 0 {
+		if w.resuming[0] != nil {
+			return w.resuming[0]
+		}
+		w.resuming = w.resuming[1:len(w.resuming)]
+	}
+	return nil
+}
+
+// dispatchEvent sends a WatchResponse to the appropriate watcher stream
+func (w *watchGrpcStream) dispatchEvent(pbresp *pb.WatchResponse) bool {
+	events := make([]*Event, len(pbresp.Events))
+	for i, ev := range pbresp.Events {
+		events[i] = (*Event)(ev)
+	}
+	wr := &WatchResponse{
+		Header:          *pbresp.Header,
+		Events:          events,
+		CompactRevision: pbresp.CompactRevision,
+		Created:         pbresp.Created,
+		Canceled:        pbresp.Canceled,
+		cancelReason:    pbresp.CancelReason,
+	}
+	ws, ok := w.substreams[pbresp.WatchId]
+	if !ok {
+		return false
+	}
+	select {
+	case ws.recvc <- wr:
+	case <-ws.donec:
+		return false
+	}
+	return true
+}
+
+// serveWatchClient forwards messages from the grpc stream to run()
+func (w *watchGrpcStream) serveWatchClient(wc pb.Watch_WatchClient) {
+	for {
+		resp, err := wc.Recv()
+		if err != nil {
+			select {
+			case w.errc <- err:
+			case <-w.donec:
+			}
+			return
+		}
+		select {
+		case w.respc <- resp:
+		case <-w.donec:
+			return
+		}
+	}
+}
+
+// serveSubstream forwards watch responses from run() to the subscriber
+func (w *watchGrpcStream) serveSubstream(ws *watcherStream, resumec chan struct{}) {
+	if ws.closing {
+		panic("created substream goroutine but substream is closing")
+	}
+
+	// nextRev is the minimum expected next revision
+	nextRev := ws.initReq.rev
+	resuming := false
+	defer func() {
+		if !resuming {
+			ws.closing = true
+		}
+		close(ws.donec)
+		if !resuming {
+			w.closingc <- ws
+		}
+		w.wg.Done()
+	}()
+
+	emptyWr := &WatchResponse{}
+	for {
+		curWr := emptyWr
+		outc := ws.outc
+
+		if len(ws.buf) > 0 {
+			curWr = ws.buf[0]
+		} else {
+			outc = nil
+		}
+		select {
+		case outc <- *curWr:
+			if ws.buf[0].Err() != nil {
+				return
+			}
+			ws.buf[0] = nil
+			ws.buf = ws.buf[1:]
+		case wr, ok := <-ws.recvc:
+			if !ok {
+				// shutdown from closeSubstream
+				return
+			}
+
+			if wr.Created {
+				if ws.initReq.retc != nil {
+					ws.initReq.retc <- ws.outc
+					// to prevent next write from taking the slot in buffered channel
+					// and posting duplicate create events
+					ws.initReq.retc = nil
+
+					// send first creation event only if requested
+					if ws.initReq.createdNotify {
+						ws.outc <- *wr
+					}
+					// once the watch channel is returned, a current revision
+					// watch must resume at the store revision. This is necessary
+					// for the following case to work as expected:
+					//	wch := m1.Watch("a")
+					//	m2.Put("a", "b")
+					//	<-wch
+					// If the revision is only bound on the first observed event,
+					// if wch is disconnected before the Put is issued, then reconnects
+					// after it is committed, it'll miss the Put.
+					if ws.initReq.rev == 0 {
+						nextRev = wr.Header.Revision
+					}
+				}
+			} else {
+				// current progress of watch; <= store revision
+				nextRev = wr.Header.Revision
+			}
+
+			if len(wr.Events) > 0 {
+				nextRev = wr.Events[len(wr.Events)-1].Kv.ModRevision + 1
+			}
+			ws.initReq.rev = nextRev
+
+			// created event is already sent above,
+			// watcher should not post duplicate events
+			if wr.Created {
+				continue
+			}
+
+			// TODO pause channel if buffer gets too large
+			ws.buf = append(ws.buf, wr)
+		case <-w.ctx.Done():
+			return
+		case <-ws.initReq.ctx.Done():
+			return
+		case <-resumec:
+			resuming = true
+			return
+		}
+	}
+	// lazily send cancel message if events on missing id
+}
+
+func (w *watchGrpcStream) newWatchClient() (pb.Watch_WatchClient, error) {
+	// mark all substreams as resuming
+	close(w.resumec)
+	w.resumec = make(chan struct{})
+	w.joinSubstreams()
+	for _, ws := range w.substreams {
+		ws.id = -1
+		w.resuming = append(w.resuming, ws)
+	}
+	// strip out nils, if any
+	var resuming []*watcherStream
+	for _, ws := range w.resuming {
+		if ws != nil {
+			resuming = append(resuming, ws)
+		}
+	}
+	w.resuming = resuming
+	w.substreams = make(map[int64]*watcherStream)
+
+	// connect to grpc stream while accepting watcher cancelation
+	stopc := make(chan struct{})
+	donec := w.waitCancelSubstreams(stopc)
+	wc, err := w.openWatchClient()
+	close(stopc)
+	<-donec
+
+	// serve all non-closing streams, even if there's a client error
+	// so that the teardown path can shutdown the streams as expected.
+	for _, ws := range w.resuming {
+		if ws.closing {
+			continue
+		}
+		ws.donec = make(chan struct{})
+		w.wg.Add(1)
+		go w.serveSubstream(ws, w.resumec)
+	}
+
+	if err != nil {
+		return nil, v3rpc.Error(err)
+	}
+
+	// receive data from new grpc stream
+	go w.serveWatchClient(wc)
+	return wc, nil
+}
+
+func (w *watchGrpcStream) waitCancelSubstreams(stopc <-chan struct{}) <-chan struct{} {
+	var wg sync.WaitGroup
+	wg.Add(len(w.resuming))
+	donec := make(chan struct{})
+	for i := range w.resuming {
+		go func(ws *watcherStream) {
+			defer wg.Done()
+			if ws.closing {
+				if ws.initReq.ctx.Err() != nil && ws.outc != nil {
+					close(ws.outc)
+					ws.outc = nil
+				}
+				return
+			}
+			select {
+			case <-ws.initReq.ctx.Done():
+				// closed ws will be removed from resuming
+				ws.closing = true
+				close(ws.outc)
+				ws.outc = nil
+				w.wg.Add(1)
+				go func() {
+					defer w.wg.Done()
+					w.closingc <- ws
+				}()
+			case <-stopc:
+			}
+		}(w.resuming[i])
+	}
+	go func() {
+		defer close(donec)
+		wg.Wait()
+	}()
+	return donec
+}
+
+// joinSubstream waits for all substream goroutines to complete
+func (w *watchGrpcStream) joinSubstreams() {
+	for _, ws := range w.substreams {
+		<-ws.donec
+	}
+	for _, ws := range w.resuming {
+		if ws != nil {
+			<-ws.donec
+		}
+	}
+}
+
+// openWatchClient retries opening a watchclient until retryConnection fails
+func (w *watchGrpcStream) openWatchClient() (ws pb.Watch_WatchClient, err error) {
+	for {
+		select {
+		case <-w.ctx.Done():
+			if err == nil {
+				return nil, w.ctx.Err()
+			}
+			return nil, err
+		default:
+		}
+		if ws, err = w.remote.Watch(w.ctx, grpc.FailFast(false)); ws != nil && err == nil {
+			break
+		}
+		if isHaltErr(w.ctx, err) {
+			return nil, v3rpc.Error(err)
+		}
+	}
+	return ws, nil
+}
+
+// toPB converts an internal watch request structure to its protobuf messagefunc (wr *watchRequest)
+func (wr *watchRequest) toPB() *pb.WatchRequest {
+	req := &pb.WatchCreateRequest{
+		StartRevision:  wr.rev,
+		Key:            []byte(wr.key),
+		RangeEnd:       []byte(wr.end),
+		ProgressNotify: wr.progressNotify,
+		Filters:        wr.filters,
+		PrevKv:         wr.prevKV,
+	}
+	cr := &pb.WatchRequest_CreateRequest{CreateRequest: req}
+	return &pb.WatchRequest{RequestUnion: cr}
+}

vendor/github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes/doc.go

@@ -0,0 +1,16 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Package rpctypes has types and values shared by the etcd server and client for v3 RPC interaction.
+package rpctypes

vendor/github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes/error.go

@@ -0,0 +1,190 @@
+// Copyright 2015 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package rpctypes
+
+import (
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+)
+
+var (
+	// server-side error
+	ErrGRPCEmptyKey      = grpc.Errorf(codes.InvalidArgument, "etcdserver: key is not provided")
+	ErrGRPCKeyNotFound   = grpc.Errorf(codes.InvalidArgument, "etcdserver: key not found")
+	ErrGRPCValueProvided = grpc.Errorf(codes.InvalidArgument, "etcdserver: value is provided")
+	ErrGRPCLeaseProvided = grpc.Errorf(codes.InvalidArgument, "etcdserver: lease is provided")
+	ErrGRPCTooManyOps    = grpc.Errorf(codes.InvalidArgument, "etcdserver: too many operations in txn request")
+	ErrGRPCDuplicateKey  = grpc.Errorf(codes.InvalidArgument, "etcdserver: duplicate key given in txn request")
+	ErrGRPCCompacted     = grpc.Errorf(codes.OutOfRange, "etcdserver: mvcc: required revision has been compacted")
+	ErrGRPCFutureRev     = grpc.Errorf(codes.OutOfRange, "etcdserver: mvcc: required revision is a future revision")
+	ErrGRPCNoSpace       = grpc.Errorf(codes.ResourceExhausted, "etcdserver: mvcc: database space exceeded")
+
+	ErrGRPCLeaseNotFound = grpc.Errorf(codes.NotFound, "etcdserver: requested lease not found")
+	ErrGRPCLeaseExist    = grpc.Errorf(codes.FailedPrecondition, "etcdserver: lease already exists")
+
+	ErrGRPCMemberExist            = grpc.Errorf(codes.FailedPrecondition, "etcdserver: member ID already exist")
+	ErrGRPCPeerURLExist           = grpc.Errorf(codes.FailedPrecondition, "etcdserver: Peer URLs already exists")
+	ErrGRPCMemberNotEnoughStarted = grpc.Errorf(codes.FailedPrecondition, "etcdserver: re-configuration failed due to not enough started members")
+	ErrGRPCMemberBadURLs          = grpc.Errorf(codes.InvalidArgument, "etcdserver: given member URLs are invalid")
+	ErrGRPCMemberNotFound         = grpc.Errorf(codes.NotFound, "etcdserver: member not found")
+
+	ErrGRPCRequestTooLarge        = grpc.Errorf(codes.InvalidArgument, "etcdserver: request is too large")
+	ErrGRPCRequestTooManyRequests = grpc.Errorf(codes.ResourceExhausted, "etcdserver: too many requests")
+
+	ErrGRPCRootUserNotExist     = grpc.Errorf(codes.FailedPrecondition, "etcdserver: root user does not exist")
+	ErrGRPCRootRoleNotExist     = grpc.Errorf(codes.FailedPrecondition, "etcdserver: root user does not have root role")
+	ErrGRPCUserAlreadyExist     = grpc.Errorf(codes.FailedPrecondition, "etcdserver: user name already exists")
+	ErrGRPCUserEmpty            = grpc.Errorf(codes.InvalidArgument, "etcdserver: user name is empty")
+	ErrGRPCUserNotFound         = grpc.Errorf(codes.FailedPrecondition, "etcdserver: user name not found")
+	ErrGRPCRoleAlreadyExist     = grpc.Errorf(codes.FailedPrecondition, "etcdserver: role name already exists")
+	ErrGRPCRoleNotFound         = grpc.Errorf(codes.FailedPrecondition, "etcdserver: role name not found")
+	ErrGRPCAuthFailed           = grpc.Errorf(codes.InvalidArgument, "etcdserver: authentication failed, invalid user ID or password")
+	ErrGRPCPermissionDenied     = grpc.Errorf(codes.PermissionDenied, "etcdserver: permission denied")
+	ErrGRPCRoleNotGranted       = grpc.Errorf(codes.FailedPrecondition, "etcdserver: role is not granted to the user")
+	ErrGRPCPermissionNotGranted = grpc.Errorf(codes.FailedPrecondition, "etcdserver: permission is not granted to the role")
+	ErrGRPCAuthNotEnabled       = grpc.Errorf(codes.FailedPrecondition, "etcdserver: authentication is not enabled")
+	ErrGRPCInvalidAuthToken     = grpc.Errorf(codes.Unauthenticated, "etcdserver: invalid auth token")
+	ErrGRPCInvalidAuthMgmt      = grpc.Errorf(codes.InvalidArgument, "etcdserver: invalid auth management")
+
+	ErrGRPCNoLeader                   = grpc.Errorf(codes.Unavailable, "etcdserver: no leader")
+	ErrGRPCNotCapable                 = grpc.Errorf(codes.Unavailable, "etcdserver: not capable")
+	ErrGRPCStopped                    = grpc.Errorf(codes.Unavailable, "etcdserver: server stopped")
+	ErrGRPCTimeout                    = grpc.Errorf(codes.Unavailable, "etcdserver: request timed out")
+	ErrGRPCTimeoutDueToLeaderFail     = grpc.Errorf(codes.Unavailable, "etcdserver: request timed out, possibly due to previous leader failure")
+	ErrGRPCTimeoutDueToConnectionLost = grpc.Errorf(codes.Unavailable, "etcdserver: request timed out, possibly due to connection lost")
+	ErrGRPCUnhealthy                  = grpc.Errorf(codes.Unavailable, "etcdserver: unhealthy cluster")
+
+	errStringToError = map[string]error{
+		grpc.ErrorDesc(ErrGRPCEmptyKey):      ErrGRPCEmptyKey,
+		grpc.ErrorDesc(ErrGRPCKeyNotFound):   ErrGRPCKeyNotFound,
+		grpc.ErrorDesc(ErrGRPCValueProvided): ErrGRPCValueProvided,
+		grpc.ErrorDesc(ErrGRPCLeaseProvided): ErrGRPCLeaseProvided,
+
+		grpc.ErrorDesc(ErrGRPCTooManyOps):   ErrGRPCTooManyOps,
+		grpc.ErrorDesc(ErrGRPCDuplicateKey): ErrGRPCDuplicateKey,
+		grpc.ErrorDesc(ErrGRPCCompacted):    ErrGRPCCompacted,
+		grpc.ErrorDesc(ErrGRPCFutureRev):    ErrGRPCFutureRev,
+		grpc.ErrorDesc(ErrGRPCNoSpace):      ErrGRPCNoSpace,
+
+		grpc.ErrorDesc(ErrGRPCLeaseNotFound): ErrGRPCLeaseNotFound,
+		grpc.ErrorDesc(ErrGRPCLeaseExist):    ErrGRPCLeaseExist,
+
+		grpc.ErrorDesc(ErrGRPCMemberExist):            ErrGRPCMemberExist,
+		grpc.ErrorDesc(ErrGRPCPeerURLExist):           ErrGRPCPeerURLExist,
+		grpc.ErrorDesc(ErrGRPCMemberNotEnoughStarted): ErrGRPCMemberNotEnoughStarted,
+		grpc.ErrorDesc(ErrGRPCMemberBadURLs):          ErrGRPCMemberBadURLs,
+		grpc.ErrorDesc(ErrGRPCMemberNotFound):         ErrGRPCMemberNotFound,
+
+		grpc.ErrorDesc(ErrGRPCRequestTooLarge):        ErrGRPCRequestTooLarge,
+		grpc.ErrorDesc(ErrGRPCRequestTooManyRequests): ErrGRPCRequestTooManyRequests,
+
+		grpc.ErrorDesc(ErrGRPCRootUserNotExist):     ErrGRPCRootUserNotExist,
+		grpc.ErrorDesc(ErrGRPCRootRoleNotExist):     ErrGRPCRootRoleNotExist,
+		grpc.ErrorDesc(ErrGRPCUserAlreadyExist):     ErrGRPCUserAlreadyExist,
+		grpc.ErrorDesc(ErrGRPCUserEmpty):            ErrGRPCUserEmpty,
+		grpc.ErrorDesc(ErrGRPCUserNotFound):         ErrGRPCUserNotFound,
+		grpc.ErrorDesc(ErrGRPCRoleAlreadyExist):     ErrGRPCRoleAlreadyExist,
+		grpc.ErrorDesc(ErrGRPCRoleNotFound):         ErrGRPCRoleNotFound,
+		grpc.ErrorDesc(ErrGRPCAuthFailed):           ErrGRPCAuthFailed,
+		grpc.ErrorDesc(ErrGRPCPermissionDenied):     ErrGRPCPermissionDenied,
+		grpc.ErrorDesc(ErrGRPCRoleNotGranted):       ErrGRPCRoleNotGranted,
+		grpc.ErrorDesc(ErrGRPCPermissionNotGranted): ErrGRPCPermissionNotGranted,
+		grpc.ErrorDesc(ErrGRPCAuthNotEnabled):       ErrGRPCAuthNotEnabled,
+		grpc.ErrorDesc(ErrGRPCInvalidAuthToken):     ErrGRPCInvalidAuthToken,
+		grpc.ErrorDesc(ErrGRPCInvalidAuthMgmt):      ErrGRPCInvalidAuthMgmt,
+
+		grpc.ErrorDesc(ErrGRPCNoLeader):                   ErrGRPCNoLeader,
+		grpc.ErrorDesc(ErrGRPCNotCapable):                 ErrGRPCNotCapable,
+		grpc.ErrorDesc(ErrGRPCStopped):                    ErrGRPCStopped,
+		grpc.ErrorDesc(ErrGRPCTimeout):                    ErrGRPCTimeout,
+		grpc.ErrorDesc(ErrGRPCTimeoutDueToLeaderFail):     ErrGRPCTimeoutDueToLeaderFail,
+		grpc.ErrorDesc(ErrGRPCTimeoutDueToConnectionLost): ErrGRPCTimeoutDueToConnectionLost,
+		grpc.ErrorDesc(ErrGRPCUnhealthy):                  ErrGRPCUnhealthy,
+	}
+
+	// client-side error
+	ErrEmptyKey      = Error(ErrGRPCEmptyKey)
+	ErrKeyNotFound   = Error(ErrGRPCKeyNotFound)
+	ErrValueProvided = Error(ErrGRPCValueProvided)
+	ErrLeaseProvided = Error(ErrGRPCLeaseProvided)
+	ErrTooManyOps    = Error(ErrGRPCTooManyOps)
+	ErrDuplicateKey  = Error(ErrGRPCDuplicateKey)
+	ErrCompacted     = Error(ErrGRPCCompacted)
+	ErrFutureRev     = Error(ErrGRPCFutureRev)
+	ErrNoSpace       = Error(ErrGRPCNoSpace)
+
+	ErrLeaseNotFound = Error(ErrGRPCLeaseNotFound)
+	ErrLeaseExist    = Error(ErrGRPCLeaseExist)
+
+	ErrMemberExist            = Error(ErrGRPCMemberExist)
+	ErrPeerURLExist           = Error(ErrGRPCPeerURLExist)
+	ErrMemberNotEnoughStarted = Error(ErrGRPCMemberNotEnoughStarted)
+	ErrMemberBadURLs          = Error(ErrGRPCMemberBadURLs)
+	ErrMemberNotFound         = Error(ErrGRPCMemberNotFound)
+
+	ErrRequestTooLarge = Error(ErrGRPCRequestTooLarge)
+	ErrTooManyRequests = Error(ErrGRPCRequestTooManyRequests)
+
+	ErrRootUserNotExist     = Error(ErrGRPCRootUserNotExist)
+	ErrRootRoleNotExist     = Error(ErrGRPCRootRoleNotExist)
+	ErrUserAlreadyExist     = Error(ErrGRPCUserAlreadyExist)
+	ErrUserEmpty            = Error(ErrGRPCUserEmpty)
+	ErrUserNotFound         = Error(ErrGRPCUserNotFound)
+	ErrRoleAlreadyExist     = Error(ErrGRPCRoleAlreadyExist)
+	ErrRoleNotFound         = Error(ErrGRPCRoleNotFound)
+	ErrAuthFailed           = Error(ErrGRPCAuthFailed)
+	ErrPermissionDenied     = Error(ErrGRPCPermissionDenied)
+	ErrRoleNotGranted       = Error(ErrGRPCRoleNotGranted)
+	ErrPermissionNotGranted = Error(ErrGRPCPermissionNotGranted)
+	ErrAuthNotEnabled       = Error(ErrGRPCAuthNotEnabled)
+	ErrInvalidAuthToken     = Error(ErrGRPCInvalidAuthToken)
+	ErrInvalidAuthMgmt      = Error(ErrGRPCInvalidAuthMgmt)
+
+	ErrNoLeader                   = Error(ErrGRPCNoLeader)
+	ErrNotCapable                 = Error(ErrGRPCNotCapable)
+	ErrStopped                    = Error(ErrGRPCStopped)
+	ErrTimeout                    = Error(ErrGRPCTimeout)
+	ErrTimeoutDueToLeaderFail     = Error(ErrGRPCTimeoutDueToLeaderFail)
+	ErrTimeoutDueToConnectionLost = Error(ErrGRPCTimeoutDueToConnectionLost)
+	ErrUnhealthy                  = Error(ErrGRPCUnhealthy)
+)
+
+// EtcdError defines gRPC server errors.
+// (https://github.com/grpc/grpc-go/blob/master/rpc_util.go#L319-L323)
+type EtcdError struct {
+	code codes.Code
+	desc string
+}
+
+// Code returns grpc/codes.Code.
+// TODO: define clientv3/codes.Code.
+func (e EtcdError) Code() codes.Code {
+	return e.code
+}
+
+func (e EtcdError) Error() string {
+	return e.desc
+}
+
+func Error(err error) error {
+	if err == nil {
+		return nil
+	}
+	verr, ok := errStringToError[grpc.ErrorDesc(err)]
+	if !ok { // not gRPC error
+		return err
+	}
+	return EtcdError{code: grpc.Code(verr), desc: grpc.ErrorDesc(verr)}
+}

vendor/github.com/coreos/etcd/etcdserver/api/v3rpc/rpctypes/md.go

@@ -0,0 +1,20 @@
+// Copyright 2016 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package rpctypes
+
+var (
+	MetadataRequireLeaderKey = "hasleader"
+	MetadataHasLeader        = "true"
+)

vendor/github.com/coreos/etcd/main.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -20,7 +20,6 @@
 // This package should NOT be extended or modified in any way; to modify the
 // etcd binary, work in the `github.com/coreos/etcd/etcdmain` package.
 //
-
 package main
 
 import "github.com/coreos/etcd/etcdmain"

vendor/github.com/coreos/etcd/mvcc/mvccpb/kv.pb.go

@@ -0,0 +1,735 @@
+// Code generated by protoc-gen-gogo.
+// source: kv.proto
+// DO NOT EDIT!
+
+/*
+	Package mvccpb is a generated protocol buffer package.
+
+	It is generated from these files:
+		kv.proto
+
+	It has these top-level messages:
+		KeyValue
+		Event
+*/
+package mvccpb
+
+import (
+	"fmt"
+
+	proto "github.com/golang/protobuf/proto"
+
+	math "math"
+
+	io "io"
+)
+
+// Reference imports to suppress errors if they are not otherwise used.
+var _ = proto.Marshal
+var _ = fmt.Errorf
+var _ = math.Inf
+
+// This is a compile-time assertion to ensure that this generated file
+// is compatible with the proto package it is being compiled against.
+// A compilation error at this line likely means your copy of the
+// proto package needs to be updated.
+const _ = proto.ProtoPackageIsVersion2 // please upgrade the proto package
+
+type Event_EventType int32
+
+const (
+	PUT    Event_EventType = 0
+	DELETE Event_EventType = 1
+)
+
+var Event_EventType_name = map[int32]string{
+	0: "PUT",
+	1: "DELETE",
+}
+var Event_EventType_value = map[string]int32{
+	"PUT":    0,
+	"DELETE": 1,
+}
+
+func (x Event_EventType) String() string {
+	return proto.EnumName(Event_EventType_name, int32(x))
+}
+func (Event_EventType) EnumDescriptor() ([]byte, []int) { return fileDescriptorKv, []int{1, 0} }
+
+type KeyValue struct {
+	// key is the key in bytes. An empty key is not allowed.
+	Key []byte `protobuf:"bytes,1,opt,name=key,proto3" json:"key,omitempty"`
+	// create_revision is the revision of last creation on this key.
+	CreateRevision int64 `protobuf:"varint,2,opt,name=create_revision,json=createRevision,proto3" json:"create_revision,omitempty"`
+	// mod_revision is the revision of last modification on this key.
+	ModRevision int64 `protobuf:"varint,3,opt,name=mod_revision,json=modRevision,proto3" json:"mod_revision,omitempty"`
+	// version is the version of the key. A deletion resets
+	// the version to zero and any modification of the key
+	// increases its version.
+	Version int64 `protobuf:"varint,4,opt,name=version,proto3" json:"version,omitempty"`
+	// value is the value held by the key, in bytes.
+	Value []byte `protobuf:"bytes,5,opt,name=value,proto3" json:"value,omitempty"`
+	// lease is the ID of the lease that attached to key.
+	// When the attached lease expires, the key will be deleted.
+	// If lease is 0, then no lease is attached to the key.
+	Lease int64 `protobuf:"varint,6,opt,name=lease,proto3" json:"lease,omitempty"`
+}
+
+func (m *KeyValue) Reset()                    { *m = KeyValue{} }
+func (m *KeyValue) String() string            { return proto.CompactTextString(m) }
+func (*KeyValue) ProtoMessage()               {}
+func (*KeyValue) Descriptor() ([]byte, []int) { return fileDescriptorKv, []int{0} }
+
+type Event struct {
+	// type is the kind of event. If type is a PUT, it indicates
+	// new data has been stored to the key. If type is a DELETE,
+	// it indicates the key was deleted.
+	Type Event_EventType `protobuf:"varint,1,opt,name=type,proto3,enum=mvccpb.Event_EventType" json:"type,omitempty"`
+	// kv holds the KeyValue for the event.
+	// A PUT event contains current kv pair.
+	// A PUT event with kv.Version=1 indicates the creation of a key.
+	// A DELETE/EXPIRE event contains the deleted key with
+	// its modification revision set to the revision of deletion.
+	Kv *KeyValue `protobuf:"bytes,2,opt,name=kv" json:"kv,omitempty"`
+	// prev_kv holds the key-value pair before the event happens.
+	PrevKv *KeyValue `protobuf:"bytes,3,opt,name=prev_kv,json=prevKv" json:"prev_kv,omitempty"`
+}
+
+func (m *Event) Reset()                    { *m = Event{} }
+func (m *Event) String() string            { return proto.CompactTextString(m) }
+func (*Event) ProtoMessage()               {}
+func (*Event) Descriptor() ([]byte, []int) { return fileDescriptorKv, []int{1} }
+
+func init() {
+	proto.RegisterType((*KeyValue)(nil), "mvccpb.KeyValue")
+	proto.RegisterType((*Event)(nil), "mvccpb.Event")
+	proto.RegisterEnum("mvccpb.Event_EventType", Event_EventType_name, Event_EventType_value)
+}
+func (m *KeyValue) Marshal() (dAtA []byte, err error) {
+	size := m.Size()
+	dAtA = make([]byte, size)
+	n, err := m.MarshalTo(dAtA)
+	if err != nil {
+		return nil, err
+	}
+	return dAtA[:n], nil
+}
+
+func (m *KeyValue) MarshalTo(dAtA []byte) (int, error) {
+	var i int
+	_ = i
+	var l int
+	_ = l
+	if len(m.Key) > 0 {
+		dAtA[i] = 0xa
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(len(m.Key)))
+		i += copy(dAtA[i:], m.Key)
+	}
+	if m.CreateRevision != 0 {
+		dAtA[i] = 0x10
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.CreateRevision))
+	}
+	if m.ModRevision != 0 {
+		dAtA[i] = 0x18
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.ModRevision))
+	}
+	if m.Version != 0 {
+		dAtA[i] = 0x20
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.Version))
+	}
+	if len(m.Value) > 0 {
+		dAtA[i] = 0x2a
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(len(m.Value)))
+		i += copy(dAtA[i:], m.Value)
+	}
+	if m.Lease != 0 {
+		dAtA[i] = 0x30
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.Lease))
+	}
+	return i, nil
+}
+
+func (m *Event) Marshal() (dAtA []byte, err error) {
+	size := m.Size()
+	dAtA = make([]byte, size)
+	n, err := m.MarshalTo(dAtA)
+	if err != nil {
+		return nil, err
+	}
+	return dAtA[:n], nil
+}
+
+func (m *Event) MarshalTo(dAtA []byte) (int, error) {
+	var i int
+	_ = i
+	var l int
+	_ = l
+	if m.Type != 0 {
+		dAtA[i] = 0x8
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.Type))
+	}
+	if m.Kv != nil {
+		dAtA[i] = 0x12
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.Kv.Size()))
+		n1, err := m.Kv.MarshalTo(dAtA[i:])
+		if err != nil {
+			return 0, err
+		}
+		i += n1
+	}
+	if m.PrevKv != nil {
+		dAtA[i] = 0x1a
+		i++
+		i = encodeVarintKv(dAtA, i, uint64(m.PrevKv.Size()))
+		n2, err := m.PrevKv.MarshalTo(dAtA[i:])
+		if err != nil {
+			return 0, err
+		}
+		i += n2
+	}
+	return i, nil
+}
+
+func encodeFixed64Kv(dAtA []byte, offset int, v uint64) int {
+	dAtA[offset] = uint8(v)
+	dAtA[offset+1] = uint8(v >> 8)
+	dAtA[offset+2] = uint8(v >> 16)
+	dAtA[offset+3] = uint8(v >> 24)
+	dAtA[offset+4] = uint8(v >> 32)
+	dAtA[offset+5] = uint8(v >> 40)
+	dAtA[offset+6] = uint8(v >> 48)
+	dAtA[offset+7] = uint8(v >> 56)
+	return offset + 8
+}
+func encodeFixed32Kv(dAtA []byte, offset int, v uint32) int {
+	dAtA[offset] = uint8(v)
+	dAtA[offset+1] = uint8(v >> 8)
+	dAtA[offset+2] = uint8(v >> 16)
+	dAtA[offset+3] = uint8(v >> 24)
+	return offset + 4
+}
+func encodeVarintKv(dAtA []byte, offset int, v uint64) int {
+	for v >= 1<<7 {
+		dAtA[offset] = uint8(v&0x7f | 0x80)
+		v >>= 7
+		offset++
+	}
+	dAtA[offset] = uint8(v)
+	return offset + 1
+}
+func (m *KeyValue) Size() (n int) {
+	var l int
+	_ = l
+	l = len(m.Key)
+	if l > 0 {
+		n += 1 + l + sovKv(uint64(l))
+	}
+	if m.CreateRevision != 0 {
+		n += 1 + sovKv(uint64(m.CreateRevision))
+	}
+	if m.ModRevision != 0 {
+		n += 1 + sovKv(uint64(m.ModRevision))
+	}
+	if m.Version != 0 {
+		n += 1 + sovKv(uint64(m.Version))
+	}
+	l = len(m.Value)
+	if l > 0 {
+		n += 1 + l + sovKv(uint64(l))
+	}
+	if m.Lease != 0 {
+		n += 1 + sovKv(uint64(m.Lease))
+	}
+	return n
+}
+
+func (m *Event) Size() (n int) {
+	var l int
+	_ = l
+	if m.Type != 0 {
+		n += 1 + sovKv(uint64(m.Type))
+	}
+	if m.Kv != nil {
+		l = m.Kv.Size()
+		n += 1 + l + sovKv(uint64(l))
+	}
+	if m.PrevKv != nil {
+		l = m.PrevKv.Size()
+		n += 1 + l + sovKv(uint64(l))
+	}
+	return n
+}
+
+func sovKv(x uint64) (n int) {
+	for {
+		n++
+		x >>= 7
+		if x == 0 {
+			break
+		}
+	}
+	return n
+}
+func sozKv(x uint64) (n int) {
+	return sovKv(uint64((x << 1) ^ uint64((int64(x) >> 63))))
+}
+func (m *KeyValue) Unmarshal(dAtA []byte) error {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		preIndex := iNdEx
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return ErrIntOverflowKv
+			}
+			if iNdEx >= l {
+				return io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		fieldNum := int32(wire >> 3)
+		wireType := int(wire & 0x7)
+		if wireType == 4 {
+			return fmt.Errorf("proto: KeyValue: wiretype end group for non-group")
+		}
+		if fieldNum <= 0 {
+			return fmt.Errorf("proto: KeyValue: illegal tag %d (wire type %d)", fieldNum, wire)
+		}
+		switch fieldNum {
+		case 1:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Key", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthKv
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Key = append(m.Key[:0], dAtA[iNdEx:postIndex]...)
+			if m.Key == nil {
+				m.Key = []byte{}
+			}
+			iNdEx = postIndex
+		case 2:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field CreateRevision", wireType)
+			}
+			m.CreateRevision = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.CreateRevision |= (int64(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		case 3:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field ModRevision", wireType)
+			}
+			m.ModRevision = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.ModRevision |= (int64(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		case 4:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Version", wireType)
+			}
+			m.Version = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.Version |= (int64(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		case 5:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Value", wireType)
+			}
+			var byteLen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				byteLen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if byteLen < 0 {
+				return ErrInvalidLengthKv
+			}
+			postIndex := iNdEx + byteLen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			m.Value = append(m.Value[:0], dAtA[iNdEx:postIndex]...)
+			if m.Value == nil {
+				m.Value = []byte{}
+			}
+			iNdEx = postIndex
+		case 6:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Lease", wireType)
+			}
+			m.Lease = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.Lease |= (int64(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		default:
+			iNdEx = preIndex
+			skippy, err := skipKv(dAtA[iNdEx:])
+			if err != nil {
+				return err
+			}
+			if skippy < 0 {
+				return ErrInvalidLengthKv
+			}
+			if (iNdEx + skippy) > l {
+				return io.ErrUnexpectedEOF
+			}
+			iNdEx += skippy
+		}
+	}
+
+	if iNdEx > l {
+		return io.ErrUnexpectedEOF
+	}
+	return nil
+}
+func (m *Event) Unmarshal(dAtA []byte) error {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		preIndex := iNdEx
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return ErrIntOverflowKv
+			}
+			if iNdEx >= l {
+				return io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		fieldNum := int32(wire >> 3)
+		wireType := int(wire & 0x7)
+		if wireType == 4 {
+			return fmt.Errorf("proto: Event: wiretype end group for non-group")
+		}
+		if fieldNum <= 0 {
+			return fmt.Errorf("proto: Event: illegal tag %d (wire type %d)", fieldNum, wire)
+		}
+		switch fieldNum {
+		case 1:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Type", wireType)
+			}
+			m.Type = 0
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				m.Type |= (Event_EventType(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+		case 2:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field Kv", wireType)
+			}
+			var msglen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				msglen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if msglen < 0 {
+				return ErrInvalidLengthKv
+			}
+			postIndex := iNdEx + msglen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			if m.Kv == nil {
+				m.Kv = &KeyValue{}
+			}
+			if err := m.Kv.Unmarshal(dAtA[iNdEx:postIndex]); err != nil {
+				return err
+			}
+			iNdEx = postIndex
+		case 3:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field PrevKv", wireType)
+			}
+			var msglen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				msglen |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if msglen < 0 {
+				return ErrInvalidLengthKv
+			}
+			postIndex := iNdEx + msglen
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			if m.PrevKv == nil {
+				m.PrevKv = &KeyValue{}
+			}
+			if err := m.PrevKv.Unmarshal(dAtA[iNdEx:postIndex]); err != nil {
+				return err
+			}
+			iNdEx = postIndex
+		default:
+			iNdEx = preIndex
+			skippy, err := skipKv(dAtA[iNdEx:])
+			if err != nil {
+				return err
+			}
+			if skippy < 0 {
+				return ErrInvalidLengthKv
+			}
+			if (iNdEx + skippy) > l {
+				return io.ErrUnexpectedEOF
+			}
+			iNdEx += skippy
+		}
+	}
+
+	if iNdEx > l {
+		return io.ErrUnexpectedEOF
+	}
+	return nil
+}
+func skipKv(dAtA []byte) (n int, err error) {
+	l := len(dAtA)
+	iNdEx := 0
+	for iNdEx < l {
+		var wire uint64
+		for shift := uint(0); ; shift += 7 {
+			if shift >= 64 {
+				return 0, ErrIntOverflowKv
+			}
+			if iNdEx >= l {
+				return 0, io.ErrUnexpectedEOF
+			}
+			b := dAtA[iNdEx]
+			iNdEx++
+			wire |= (uint64(b) & 0x7F) << shift
+			if b < 0x80 {
+				break
+			}
+		}
+		wireType := int(wire & 0x7)
+		switch wireType {
+		case 0:
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return 0, ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return 0, io.ErrUnexpectedEOF
+				}
+				iNdEx++
+				if dAtA[iNdEx-1] < 0x80 {
+					break
+				}
+			}
+			return iNdEx, nil
+		case 1:
+			iNdEx += 8
+			return iNdEx, nil
+		case 2:
+			var length int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return 0, ErrIntOverflowKv
+				}
+				if iNdEx >= l {
+					return 0, io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				length |= (int(b) & 0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			iNdEx += length
+			if length < 0 {
+				return 0, ErrInvalidLengthKv
+			}
+			return iNdEx, nil
+		case 3:
+			for {
+				var innerWire uint64
+				var start int = iNdEx
+				for shift := uint(0); ; shift += 7 {
+					if shift >= 64 {
+						return 0, ErrIntOverflowKv
+					}
+					if iNdEx >= l {
+						return 0, io.ErrUnexpectedEOF
+					}
+					b := dAtA[iNdEx]
+					iNdEx++
+					innerWire |= (uint64(b) & 0x7F) << shift
+					if b < 0x80 {
+						break
+					}
+				}
+				innerWireType := int(innerWire & 0x7)
+				if innerWireType == 4 {
+					break
+				}
+				next, err := skipKv(dAtA[start:])
+				if err != nil {
+					return 0, err
+				}
+				iNdEx = start + next
+			}
+			return iNdEx, nil
+		case 4:
+			return iNdEx, nil
+		case 5:
+			iNdEx += 4
+			return iNdEx, nil
+		default:
+			return 0, fmt.Errorf("proto: illegal wireType %d", wireType)
+		}
+	}
+	panic("unreachable")
+}
+
+var (
+	ErrInvalidLengthKv = fmt.Errorf("proto: negative length found during unmarshaling")
+	ErrIntOverflowKv   = fmt.Errorf("proto: integer overflow")
+)
+
+func init() { proto.RegisterFile("kv.proto", fileDescriptorKv) }
+
+var fileDescriptorKv = []byte{
+	// 303 bytes of a gzipped FileDescriptorProto
+	0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0x6c, 0x90, 0x41, 0x4e, 0xc2, 0x40,
+	0x14, 0x86, 0x3b, 0x14, 0x0a, 0x3e, 0x08, 0x36, 0x13, 0x12, 0x27, 0x2e, 0x26, 0x95, 0x8d, 0x18,
+	0x13, 0x4c, 0xf0, 0x06, 0xc6, 0xae, 0x70, 0x61, 0x1a, 0x74, 0x4b, 0x4a, 0x79, 0x21, 0xa4, 0x94,
+	0x69, 0x4a, 0x9d, 0xa4, 0x37, 0x71, 0xef, 0xde, 0x73, 0xb0, 0xe4, 0x08, 0x52, 0x2f, 0x62, 0xfa,
+	0xc6, 0xe2, 0xc6, 0xcd, 0xe4, 0xfd, 0xff, 0xff, 0x65, 0xe6, 0x7f, 0x03, 0x9d, 0x58, 0x8f, 0xd3,
+	0x4c, 0xe5, 0x8a, 0x3b, 0x89, 0x8e, 0xa2, 0x74, 0x71, 0x39, 0x58, 0xa9, 0x95, 0x22, 0xeb, 0xae,
+	0x9a, 0x4c, 0x3a, 0xfc, 0x64, 0xd0, 0x99, 0x62, 0xf1, 0x1a, 0x6e, 0xde, 0x90, 0xbb, 0x60, 0xc7,
+	0x58, 0x08, 0xe6, 0xb1, 0x51, 0x2f, 0xa8, 0x46, 0x7e, 0x0d, 0xe7, 0x51, 0x86, 0x61, 0x8e, 0xf3,
+	0x0c, 0xf5, 0x7a, 0xb7, 0x56, 0x5b, 0xd1, 0xf0, 0xd8, 0xc8, 0x0e, 0xfa, 0xc6, 0x0e, 0x7e, 0x5d,
+	0x7e, 0x05, 0xbd, 0x44, 0x2d, 0xff, 0x28, 0x9b, 0xa8, 0x6e, 0xa2, 0x96, 0x27, 0x44, 0x40, 0x5b,
+	0x63, 0x46, 0x69, 0x93, 0xd2, 0x5a, 0xf2, 0x01, 0xb4, 0x74, 0x55, 0x40, 0xb4, 0xe8, 0x65, 0x23,
+	0x2a, 0x77, 0x83, 0xe1, 0x0e, 0x85, 0x43, 0xb4, 0x11, 0xc3, 0x0f, 0x06, 0x2d, 0x5f, 0xe3, 0x36,
+	0xe7, 0xb7, 0xd0, 0xcc, 0x8b, 0x14, 0xa9, 0x6e, 0x7f, 0x72, 0x31, 0x36, 0x7b, 0x8e, 0x29, 0x34,
+	0xe7, 0xac, 0x48, 0x31, 0x20, 0x88, 0x7b, 0xd0, 0x88, 0x35, 0x75, 0xef, 0x4e, 0xdc, 0x1a, 0xad,
+	0x17, 0x0f, 0x1a, 0xb1, 0xe6, 0x37, 0xd0, 0x4e, 0x33, 0xd4, 0xf3, 0x58, 0x53, 0xf9, 0xff, 0x30,
+	0xa7, 0x02, 0xa6, 0x7a, 0xe8, 0xc1, 0xd9, 0xe9, 0x7e, 0xde, 0x06, 0xfb, 0xf9, 0x65, 0xe6, 0x5a,
+	0x1c, 0xc0, 0x79, 0xf4, 0x9f, 0xfc, 0x99, 0xef, 0xb2, 0x07, 0xb1, 0x3f, 0x4a, 0xeb, 0x70, 0x94,
+	0xd6, 0xbe, 0x94, 0xec, 0x50, 0x4a, 0xf6, 0x55, 0x4a, 0xf6, 0xfe, 0x2d, 0xad, 0x85, 0x43, 0xff,
+	0x7e, 0xff, 0x13, 0x00, 0x00, 0xff, 0xff, 0xb5, 0x45, 0x92, 0x5d, 0xa1, 0x01, 0x00, 0x00,
+}

vendor/github.com/coreos/etcd/pkg/srv/srv.go

@@ -0,0 +1,140 @@
+// Copyright 2015 The etcd Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Package srv looks up DNS SRV records.
+package srv
+
+import (
+	"fmt"
+	"net"
+	"net/url"
+	"strings"
+
+	"github.com/coreos/etcd/pkg/types"
+)
+
+var (
+	// indirection for testing
+	lookupSRV      = net.LookupSRV // net.DefaultResolver.LookupSRV when ctxs don't conflict
+	resolveTCPAddr = net.ResolveTCPAddr
+)
+
+// GetCluster gets the cluster information via DNS discovery.
+// Also sees each entry as a separate instance.
+func GetCluster(service, name, dns string, apurls types.URLs) ([]string, error) {
+	tempName := int(0)
+	tcp2ap := make(map[string]url.URL)
+
+	// First, resolve the apurls
+	for _, url := range apurls {
+		tcpAddr, err := resolveTCPAddr("tcp", url.Host)
+		if err != nil {
+			return nil, err
+		}
+		tcp2ap[tcpAddr.String()] = url
+	}
+
+	stringParts := []string{}
+	updateNodeMap := func(service, scheme string) error {
+		_, addrs, err := lookupSRV(service, "tcp", dns)
+		if err != nil {
+			return err
+		}
+		for _, srv := range addrs {
+			port := fmt.Sprintf("%d", srv.Port)
+			host := net.JoinHostPort(srv.Target, port)
+			tcpAddr, terr := resolveTCPAddr("tcp", host)
+			if terr != nil {
+				err = terr
+				continue
+			}
+			n := ""
+			url, ok := tcp2ap[tcpAddr.String()]
+			if ok {
+				n = name
+			}
+			if n == "" {
+				n = fmt.Sprintf("%d", tempName)
+				tempName++
+			}
+			// SRV records have a trailing dot but URL shouldn't.
+			shortHost := strings.TrimSuffix(srv.Target, ".")
+			urlHost := net.JoinHostPort(shortHost, port)
+			stringParts = append(stringParts, fmt.Sprintf("%s=%s://%s", n, scheme, urlHost))
+			if ok && url.Scheme != scheme {
+				err = fmt.Errorf("bootstrap at %s from DNS for %s has scheme mismatch with expected peer %s", scheme+"://"+urlHost, service, url.String())
+			}
+		}
+		if len(stringParts) == 0 {
+			return err
+		}
+		return nil
+	}
+
+	failCount := 0
+	err := updateNodeMap(service+"-ssl", "https")
+	srvErr := make([]string, 2)
+	if err != nil {
+		srvErr[0] = fmt.Sprintf("error querying DNS SRV records for _%s-ssl %s", service, err)
+		failCount++
+	}
+	err = updateNodeMap(service, "http")
+	if err != nil {
+		srvErr[1] = fmt.Sprintf("error querying DNS SRV records for _%s %s", service, err)
+		failCount++
+	}
+	if failCount == 2 {
+		return nil, fmt.Errorf("srv: too many errors querying DNS SRV records (%q, %q)", srvErr[0], srvErr[1])
+	}
+	return stringParts, nil
+}
+
+type SRVClients struct {
+	Endpoints []string
+	SRVs      []*net.SRV
+}
+
+// GetClient looks up the client endpoints for a service and domain.
+func GetClient(service, domain string) (*SRVClients, error) {
+	var urls []*url.URL
+	var srvs []*net.SRV
+
+	updateURLs := func(service, scheme string) error {
+		_, addrs, err := lookupSRV(service, "tcp", domain)
+		if err != nil {
+			return err
+		}
+		for _, srv := range addrs {
+			urls = append(urls, &url.URL{
+				Scheme: scheme,
+				Host:   net.JoinHostPort(srv.Target, fmt.Sprintf("%d", srv.Port)),
+			})
+		}
+		srvs = append(srvs, addrs...)
+		return nil
+	}
+
+	errHTTPS := updateURLs(service+"-ssl", "https")
+	errHTTP := updateURLs(service, "http")
+
+	if errHTTPS != nil && errHTTP != nil {
+		return nil, fmt.Errorf("dns lookup errors: %s and %s", errHTTPS, errHTTP)
+	}
+
+	endpoints := make([]string, len(urls))
+	for i := range urls {
+		endpoints[i] = urls[i].String()
+	}
+	return &SRVClients{Endpoints: endpoints, SRVs: srvs}, nil
+}

vendor/github.com/coreos/etcd/pkg/types/doc.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.

vendor/github.com/coreos/etcd/pkg/types/id.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.

vendor/github.com/coreos/etcd/pkg/types/set.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.

vendor/github.com/coreos/etcd/pkg/types/slice.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.

vendor/github.com/coreos/etcd/pkg/types/urls.go

@@ -1,4 +1,4 @@
-// Copyright 2015 CoreOS, Inc.
+// Copyright 2015 The etcd Authors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -36,8 +36,8 @@ func NewURLs(strs []string) (URLs, error) {
 		if err != nil {
 			return nil, err
 		}
-		if u.Scheme != "http" && u.Scheme != "https" {
+		if u.Scheme != "http" && u.Scheme != "https" && u.Scheme != "unix" && u.Scheme != "unixs" {
-			return nil, fmt.Errorf("URL scheme must be http or https: %s", in)
+			return nil, fmt.Errorf("URL scheme must be http, https, unix, or unixs: %s", in)
 		}
 		if _, _, err := net.SplitHostPort(u.Host); err != nil {
 			return nil, fmt.Errorf(`URL address does not have the form "host:port": %s`, in)
@@ -53,6 +53,14 @@ func NewURLs(strs []string) (URLs, error) {
 	return us, nil
 }
 
+func MustNewURLs(strs []string) URLs {
+	urls, err := NewURLs(strs)
+	if err != nil {
+		panic(err)
+	}
+	return urls
+}
+
 func (us URLs) String() string {
 	return strings.Join(us.StringSlice(), ",")
 }