将有问题的etcd节点重新加入集群

etcd集群中某个节点坏掉，或不小心清空了某个节点的数据，可以按照如下步骤重新将此节点加入集群

1.从集群中删除坏掉的节点

1.查看集群健康状态，找到坏掉的节点ID：

etcdctl --endpoints=https://10.4.63.161:2379 --ca-file=/etc/ssl/etcd/ssl/ca.pem --cert-file=/etc/ssl/etcd/ssl/node-k8smaster02.pem --key-file=/etc/ssl/etcd/ssl/node-k8smaster02-key.pem  cluster-health

2.将此节点从集群中移除

etcdctl --endpoints=https://10.4.63.161:2379 --ca-file=/etc/ssl/etcd/ssl/ca.pem --cert-file=/etc/ssl/etcd/ssl/node-k8smaster02.pem --key-file=/etc/ssl/etcd/ssl/node-k8smaster02-key.pem  member remove 6284d00a680a6741

2. 修改坏掉节点的配置文件

vim /etc/etcd.env

ETCD_DATA_DIR=/dcos/etcd
ETCD_ADVERTISE_CLIENT_URLS=https://10.4.63.160:2379
ETCD_INITIAL_ADVERTISE_PEER_URLS=https://10.4.63.160:2380
ETCD_INITIAL_CLUSTER_STATE=existing                           ------------------------修改,从new改为existing
ETCD_LISTEN_CLIENT_URLS=https://10.4.63.160:2379,https://127.0.0.1:2379
ETCD_ELECTION_TIMEOUT=5000
ETCD_HEARTBEAT_INTERVAL=250
ETCD_INITIAL_CLUSTER_TOKEN=k8s_etcd
ETCD_LISTEN_PEER_URLS=https://10.4.63.160:2380
ETCD_NAME=etcd1
ETCD_PROXY=off
ETCD_INITIAL_CLUSTER=etcd1=https://10.4.63.160:2380,etcd2=https://10.4.63.161:2380,etcd3=https://10.4.63.162:2380

# TLS settings
ETCD_TRUSTED_CA_FILE=/etc/ssl/etcd/ssl/ca.pem
ETCD_CERT_FILE=/etc/ssl/etcd/ssl/member-k8smaster01.pem
ETCD_KEY_FILE=/etc/ssl/etcd/ssl/member-k8smaster01-key.pem
ETCD_PEER_TRUSTED_CA_FILE=/etc/ssl/etcd/ssl/ca.pem
ETCD_PEER_CERT_FILE=/etc/ssl/etcd/ssl/member-k8smaster01.pem
ETCD_PEER_KEY_FILE=/etc/ssl/etcd/ssl/member-k8smaster01-key.pem
ETCD_PEER_CLIENT_CERT_AUTH=true

3.将此节点作为新成员添加入集群

etcdctl --endpoints=https://10.4.63.161:2379 --ca-file=/etc/ssl/etcd/ssl/ca.pem --cert-file=/etc/ssl/etcd/ssl/node-k8smaster02.pem --key-file=/etc/ssl/etcd/ssl/node-k8smaster02-key.pem  member add etcd1 https://10.4.63.160:2380

4.重启坏掉的节点

systemctl restart etcd