title: 部署高可用consul集群 #标题tags: consul #标签
date: 2020-12-29
categories: linux大杂烩 # 分类
记录下conusl高可用集群部署方案。
参考官方文档。
环境准备
| OS | hostname | IP | roles |
|---|---|---|---|
| Centos 7.5 | consul01 | 192.168.20.2 | server |
| Centos 7.5 | consul02 | 192.168.20.3 | server |
| Centos 7.5 | consul03 | 192.168.20.4 | server |
| Centos 7.5 | consul04 | 192.168.20.5 | client |
部署consul集群
配置consul01
# 下载consul二进制包$ wget https://releases.hashicorp.com/consul/1.9.0/consul_1.9.0_linux_amd64.zip# 创建consul家目录$ mkdir /opt/consul/{bin,conf,data,logs} -p# 解压二进制文件到指定目录$ unzip consul_1.9.0_linux_amd64.zip -d /opt/consul/bin/# 将consul家目录发送至其他节点$ for i in 3 4 5;do rsync -az /opt/consul 192.168.20.$i:/opt/;done# 准备配置文件$ cat > /opt/consul/conf/consul.json << 'EOF'{"ports": {"http": 8500,"dns": 8600,"serf_lan": 8301,"serf_wan": 8302,"server": 8300},"datacenter": "aspire-pro","data_dir": "/opt/consul/data","log_level": "INFO","log_file": "/opt/consul/logs/","node_name": "server-20-2","server": true,"ui": true,"bind_addr": "192.168.20.2","client_addr": "0.0.0.0"}EOF# 启动consul01$ nohup /opt/consul/bin/consul agent -server -bootstrap -config-dir /opt/consul/conf/ &> /dev/null &# -server：以server身份启动，默认是client。# -bootstrap：仅在第一个节点第一次启动时需要增加 -bootstrap 选项，以便引导集群初始化。# 后续任何节点启动，都不要加 -bootstrap 选项$ ss -lnpt | grep consul # 确定端口已监听LISTEN 0 128 192.168.20.2:8300 *:* users:(("consul",pid=31879,fd=7))LISTEN 0 128 192.168.20.2:8301 *:* users:(("consul",pid=31879,fd=13))LISTEN 0 128 192.168.20.2:8302 *:* users:(("consul",pid=31879,fd=9))LISTEN 0 128 [::]:8500 [::]:* users:(("consul",pid=31879,fd=17))LISTEN 0 128 [::]:8600 [::]:* users:(("consul",pid=31879,fd=16))
配置文件说明
下面是常用的配置参数解释：
- ports：指定各种通讯端口；
- datacenter：指定数据中心名称，自定义即可，默认为dc1；
- data_dir：数据存放目录；
- log_level：日志等级；
- log_file：日志存放路径，日志名为 consul-时间戳.log 命名，通常是服务重启后或者24小时会生成新的日志文件；
- node_name：此节点在ui界面中显示的名称；
- server：是否为服务端，默认为false，一个高可用集群，至少有三个节点是服务端；
- ui：是否开启此节点的ui界面，一般server端建议开启；
- bind_addr：内部群集通信应绑定到的地址。这个IP地址应该可以被集群中的所有其他节点访问。默认情况下，这是“0.0.0.0”；
- client_addr：Consul将绑定客户端接口(包括HTTP和DNS服务器)到的地址。默认情况下，这是”127.0.0.1”。
配置consul02
$ cat > /opt/consul/conf/consul.json << 'EOF'
{
"ports": {
"http": 8500,
"dns": 8600,
"serf_lan": 8301,
"serf_wan": 8302,
"server": 8300
},
"datacenter": "aspire-pro",
"data_dir": "/opt/consul/data",
"log_level": "INFO",
"log_file": "/opt/consul/logs/",
"node_name": "server-20-3",
"server": true,
"ui": true,
"bind_addr": "192.168.20.3",
"client_addr": "0.0.0.0"
}
EOF
# 启动consul
$ nohup /opt/consul/bin/consul agent -server -config-dir /opt/consul/conf/ &> /dev/null &
# 加入consul01集群
$ /opt/consul/bin/consul join 192.168.20.2
配置consul03
$ cat > /opt/consul/conf/consul.json << 'EOF'
{
"ports": {
"http": 8500,
"dns": 8600,
"serf_lan": 8301,
"serf_wan": 8302,
"server": 8300
},
"datacenter": "aspire-pro",
"data_dir": "/opt/consul/data",
"log_level": "INFO",
"log_file": "/opt/consul/logs/",
"node_name": "server-20-4",
"server": true,
"ui": true,
"bind_addr": "192.168.20.4",
"client_addr": "0.0.0.0"
}
EOF
# 启动consul
$ nohup /opt/consul/bin/consul agent -server -config-dir /opt/consul/conf/ &> /dev/null &
# 加入consul01集群
$ /opt/consul/bin/consul join 192.168.20.2
配置consul04
$ cat > /opt/consul/conf/consul.json << 'EOF'
{
"ports": {
"http": 8500,
"dns": 8600,
"serf_lan": 8301,
"serf_wan": 8302,
"server": 8300
},
"datacenter": "aspire-pro",
"data_dir": "/opt/consul/data",
"log_level": "INFO",
"log_file": "/opt/consul/logs/",
"node_name": "client-20-5",
"bind_addr": "192.168.20.5",
"client_addr": "0.0.0.0",
"retry_join": ["192.168.20.2","192.168.20.3","192.168.20.4"],
"retry_interval": "30s",
"rejoin_after_leave": true
}
EOF
# 启动consul即可
$ nohup /opt/consul/bin/consul agent -config-dir /opt/consul/conf/ &> /dev/null &
客户端配置文件中部分参数解释：
- retry_join：类似于-join，但是允许重试一个连接直到成功。一旦成功地加入到成员列表中的成员，它就不再尝试再次加入。
- retry_interval：连接尝试之间的等待时间。默认为30秒。
- rejoin_after_leave：允许使用以前的状态重新加入集群。
- start_join：指定启动时要加入的集群IP列表
注：其他更多client要加入集群的话，则按照 consul04 这样配置、启动即可。
访问consul管理页面
访问任意一个开启ui的server端的 8500端口，按照如下点击，即可看到集群中的所有节点：
重启consul01
由于consul01 节点启动时，增加了 -bootstrap 参数，故最好在集群配置成功后，重启consul01节点，正好也可以验证下consul高可用性，请在进行以下操作前，注意当前集群中的leader角色在哪个节点（一般是在consul01节点上，因为此节点是第一个启动的，并且有-bootstrap这个参数进行集群初始化引导）。
# 停止consul（在consul01节点执行）
$ ps -ef | grep consul | grep -v grep | awk '{print $2}' | xargs kill -9
# 启动consul01（注意：没有 -bootstrap 参数，并且后续重启都应是下面的命令）
$ nohup /opt/consul/bin/consul agent -server -config-dir /opt/consul/conf/ &> /dev/null &
再次访问ui页面，查看当前集群角色状态如下：
consul运维指令
$ consul operator raft list-peers # 查看集群状态及server的角色
Node ID Address State Voter RaftProtocol
server-20-3 47c0d52a-cd2e-3f42-2320-00f6f230c71f 192.168.20.3:8300 leader true 3
server-20-4 a6e3102a-a3ac-ad5c-e960-6f6bd146e3c4 192.168.20.4:8300 follower true 3
server-20-2 f6d8c628-47e9-7fab-bec5-6896aa7f3fd3 192.168.20.2:8300 follower true 3
$ consul members # 查看集群和节点状态
Node Address Status Type Build Protocol DC Segment
server-20-2 192.168.20.2:8301 alive server 1.9.0 2 aspire-pro <all>
server-20-3 192.168.20.3:8301 alive server 1.9.0 2 aspire-pro <all>
server-20-4 192.168.20.4:8301 alive server 1.9.0 2 aspire-pro <all>
client-20-5 192.168.20.5:8301 alive client 1.9.0 2 aspire-pro <default>
# 集群get/set 测试
$ consul kv put key value # set一个键值
Success! Data written to: key
$ consul kv get key # get一个键值
value
$ consul info # 查询当前节点详细信息
agent:
check_monitors = 0
check_ttls = 0
checks = 0
services = 0
build:
prerelease =
revision = a417fe51
version = 1.9.0
consul:
acl = disabled
bootstrap = false
known_datacenters = 1
leader = false
leader_addr = 192.168.20.3:8300
server = true
raft:
applied_index = 458
# ....... 省略部分输出
集群中移除指定节点
$ consul operator raft list-peers # 假设集群中server节点如下
Node ID Address State Voter RaftProtocol
server-20-3 47c0d52a-cd2e-3f42-2320-00f6f230c71f 192.168.20.3:8300 leader true 3
server-20-4 a6e3102a-a3ac-ad5c-e960-6f6bd146e3c4 192.168.20.4:8300 follower true 3
server-20-2 f6d8c628-47e9-7fab-bec5-6896aa7f3fd3 192.168.20.2:8300 follower true 3
$ consul operator raft remove-peer -id=47c0d52a-cd2e-3f42-2320-00f6f230c71f # 移除当前leader角色的节点
Removed peer with id "47c0d52a-cd2e-3f42-2320-00f6f230c71f"
$ consul operator raft list-peers # 再次查看，发现并不是真正的移除，只是leader角色转移了，并且 voter 状态为 false
Node ID Address State Voter RaftProtocol
server-20-4 a6e3102a-a3ac-ad5c-e960-6f6bd146e3c4 192.168.20.4:8300 follower true 3
server-20-2 f6d8c628-47e9-7fab-bec5-6896aa7f3fd3 192.168.20.2:8300 leader true 3
server-20-3 47c0d52a-cd2e-3f42-2320-00f6f230c71f 192.168.20.3:8300 follower false 3
$ consul operator raft remove-peer -id=47c0d52a-cd2e-3f42-2320-00f6f230c71f # 再执行一次移除操作
Removed peer with id "47c0d52a-cd2e-3f42-2320-00f6f230c71f"
$ consul operator raft list-peers # 再次查看集群中的server，发现已经被移除
Node ID Address State Voter RaftProtocol
server-20-4 a6e3102a-a3ac-ad5c-e960-6f6bd146e3c4 192.168.20.4:8300 follower true 3
server-20-2 f6d8c628-47e9-7fab-bec5-6896aa7f3fd3 192.168.20.2:8300 leader true 3
若有收获，就点个赞吧
0 人点赞
