docker etcd

etcd是CoreOS团队于2013年6月发起的开源项目，它的目标是构建一个高可用的分布式键值(key-value)数据库，用于配置共享和服务发现

etcd内部采用raft协议作为一致性算法，etcd基于Go语言实现。

etcd作为服务发现系统，有以下的特点：

• 简单：安装配置简单，HTTP+JSON的 API进行交互，使用curl命令可以轻松使用
• 安全：支持SSL证书验证机制
• 快速：根据官方提供的benchmark数据，单实例支持每秒1k+写操作
• 可靠：采用raft算法，实现分布式系统数据的可用性和一致性
• 分布式系统中数据分为控制数据和应用数据，etcd处理的数据为控制数据

同一个分布式集群中的进程或服务，互相感知并建立链接，这就是服务发现。解决服务发现问题的三大支柱：

• 一个强一致性、高可用的服务存储目录
  • 基于Raft算法
• 一种注册服务和监控服务健康状态的机制
• 一种查找和链接服务的机制
• 微服务协同工作架构中，服务动态添加　

etcd基本命令

集群搭建参考k8s集群搭建里的etcd

更新一个节点

#首先查看所有节点
[root@slave2 ~]# etcdctl member list
646ec025a72fc8a: name=etcd3 peerURLs=http://192.168.132.133:2380 clientURLs=http://192.168.132.133:2379 isLeader=true
dde447f371b55f50: name=etcd1 peerURLs=http://192.168.132.131:2380 clientURLs=http://192.168.132.131:2379 isLeader=false
e96057c7f8ba5f4b: name=etcd2 peerURLs=http://192.168.132.132:2380 clientURLs=http://192.168.132.132:2379 isLeader=false

#更新一个节点
[root@slave2 ~]# etcdctl member update 646ec025a72fc8a http://192.168.132.133:2380
Updated member with ID 646ec025a72fc8a in cluster

删除一个节点

#查看所有节点
[root@slave2 ~]# etcdctl member list
646ec025a72fc8a: name=etcd3 peerURLs=http://192.168.132.133:2380 clientURLs=http://192.168.132.133:2379 isLeader=false
dde447f371b55f50: name=etcd1 peerURLs=http://192.168.132.131:2380 clientURLs=http://192.168.132.131:2379 isLeader=true
e96057c7f8ba5f4b: name=etcd2 peerURLs=http://192.168.132.132:2380 clientURLs=http://192.168.132.132:2379 isLeader=false

#删除一个节点
[root@slave2 ~]# etcdctl member remove 646ec025a72fc8a
Removed member 646ec025a72fc8a from cluster

#再次查看节点，使用当前命令报错
[root@slave2 ~]# etcdctl member list
Error:  client: etcd cluster is unavailable or misconfigured; error #0: dial tcp 127.0.0.1:2379: getsockopt: connection refused
; error #1: dial tcp 127.0.0.1:4001: getsockopt: connection refused

error #0: dial tcp 127.0.0.1:2379: getsockopt: connection refused
error #1: dial tcp 127.0.0.1:4001: getsockopt: connection refused

#换一个命令
[root@slave2 ~]#  etcdctl --endpoints "http://192.168.132.131:2379" member list
dde447f371b55f50: name=etcd1 peerURLs=http://192.168.132.131:2380 clientURLs=http://192.168.132.131:2379 isLeader=true
e96057c7f8ba5f4b: name=etcd2 peerURLs=http://192.168.132.132:2380 clientURLs=http://192.168.132.132:2379 isLeader=false

添加节点

#将删除的节点添加回来
[root@slave2 ~]# etcdctl --endpoints "http://192.168.132.132:2379" member add etcd3 http://192.168.132.133:2380
Added member named etcd3 with ID 7a6809311b4a3579 to cluster

ETCD_NAME="etcd3"
ETCD_INITIAL_CLUSTER="etcd3=http://192.168.132.133:2380,etcd1=http://192.168.132.131:2380,etcd2=http://192.168.132.132:2380"
ETCD_INITIAL_CLUSTER_STATE="existing"
#再次查看，状态为unstarted
[root@slave2 ~]# etcdctl --endpoints "http://192.168.132.132:2379" member list
7a6809311b4a3579[unstarted]: peerURLs=http://192.168.132.133:2380
dde447f371b55f50: name=etcd1 peerURLs=http://192.168.132.131:2380 clientURLs=http://192.168.132.131:2379 isLeader=true
e96057c7f8ba5f4b: name=etcd2 peerURLs=http://192.168.132.132:2380 clientURLs=http://192.168.132.132:2379 isLeader=false

清空目标节点数据

目标节点从集群中删除后，成员信息会更新。新节点是作为一个全新的节点加入集群，如果data-dir有数据，etcd启动时会读取己经存在的数据，仍然用旧的memberID会造成无法加入集群，所以一定要清空新节点的data-dir。

#删除的节点的主机
$ rm -rf /var/lib/etcd/etcd3

在目标节点上启动新增加的成员

修改配置文件中ETCD_INITIAL_CLUSTER_STATE标记为existing，如果为new，则会自动生成一个新的memberID，这和前面添加节点时生成的ID不一致，故日志中会报节点ID不匹配的错。

[root@slave2 ~]# cat /etc/etcd/etcd.conf |grep -v "^#"
[Member]
ETCD_DATA_DIR="/var/lib/etcd/etcd3"
ETCD_LISTEN_PEER_URLS="http://192.168.132.133:2380"
ETCD_LISTEN_CLIENT_URLS="http://192.168.132.133:2379,http://127.0.0.1:2379"
ETCD_NAME="etcd3"
[Clustering]
ETCD_INITIAL_ADVERTISE_PEER_URLS="http://192.168.132.133:2380"
ETCD_ADVERTISE_CLIENT_URLS="http://192.168.132.133:2379"
ETCD_INITIAL_CLUSTER_STATE="existing"
ETCD_INITIAL_CLUSTER="etcd1=http://192.168.132.131:2380,etcd2=http://192.168.132.132:2380,etcd3=http://192.168.132.133:2380"

重启服务，再次查看

[root@slave2 ~]# systemctl restart etcd
[root@slave2 ~]# etcdctl --endpoints "http://192.168.132.132:2379" member list
7a6809311b4a3579: name=etcd3 peerURLs=http://192.168.132.133:2380 clientURLs=http://192.168.132.133:2379 isLeader=false
dde447f371b55f50: name=etcd1 peerURLs=http://192.168.132.131:2380 clientURLs=http://192.168.132.131:2379 isLeader=true
e96057c7f8ba5f4b: name=etcd2 peerURLs=http://192.168.132.132:2380 clientURLs=http://192.168.132.132:2379 isLeader=false
[root@slave2 ~]# etcdctl member list
7a6809311b4a3579: name=etcd3 peerURLs=http://192.168.132.133:2380 clientURLs=http://192.168.132.133:2379 isLeader=false
dde447f371b55f50: name=etcd1 peerURLs=http://192.168.132.131:2380 clientURLs=http://192.168.132.131:2379 isLeader=true
e96057c7f8ba5f4b: name=etcd2 peerURLs=http://192.168.132.132:2380 clientURLs=http://192.168.132.132:2379 isLeader=false