Elasticsearch集群缩容

1、关闭节点分配，使上面的shard强制迁移到其他节点上

curl -XPUT localhost:9200/_cluster/settings -H 'Content-Type: application/json' -d '{
  "transient" :{
      "cluster.routing.allocation.exclude._ip" : "node_ip1,node_ip2"
   }
}'
等待shard reroute完成后再继续下一步

2、在剩余节点中将配置discovery.seed_hosts和 cluster.initial_master_nodes 中删掉要下线节点ip。不需要重启es进程

3、关机下线。注意zabbix和salt里要删掉

备注
# 更改cluster_concurrent_rebalance、node_concurrent_recoveries数量来加快分片的转移：注意集群负载
curl -X PUT "localhost:9200/_cluster/settings?pretty" -H 'Content-Type: application/json' -d'
{
  "transient": {
    "cluster.routing.allocation.cluster_concurrent_rebalance": 15,
    "cluster.routing.allocation.node_concurrent_recoveries": 15
  }
}'

# 确定集群shard reroute是否完成
#查看是否有正在RELOCATING分片
curl -s "http://localhost:9200/_cat/shards" | grep RELOCATING
#查看集群健康状态
curl http://localhost:9200/_cluster/health?pretty

#查看节点上indices信息
curl http://localhost:9200/_nodes/‘被移除节点’/stats/indices?pretty

3、最后将刚开始添加的配置修改为默认