kubernetes 二进制安装 遇到 etcd 不能启动报错 处理

查看系统日志 报错信息

May 13 14:14:02 master etcd: health check for peer 92e46651355a9596 could not connect: dial tcp 192.168.243.150:2380: connect: connection refused
May 13 14:14:02 master etcd: health check for peer 92e46651355a9596 could not connect: dial tcp 192.168.243.150:2380: connect: connection refused
May 13 14:14:02 master etcd: health check for peer df28de52717ead9b could not connect: dial tcp 192.168.243.149:2380: connect: connection refused
May 13 14:14:02 master etcd: health check for peer df28de52717ead9b could not connect: dial tcp 192.168.243.149:2380: connect: connection refused

// 1·排除 etcd 配置文件错误。
// 2. 需要先启动 其他节点的 etcd 而不是 master 的etcd
// 3.原因：分析是因为etcd1的配置文件/etc/systemd/system/etcd.service 启动脚本中的ETCD_INITIAL_CLUSTER_STATE是new，而在配置中ETCD_INITIAL_CLUSTER写入了etcd2/3的IP:PORT，这时etcd1尝试去连接etcd2、etcd3，但是etcd2、3的etcd服务此时还未启动，因此需要先启动etcd2和3的etcd服务，再去启动etcd1。