es集群快速恢复（优化方案）_大数据_ClearloveXXX的博客-CSDN博客

1、停止数据写入

 

2、关闭allocate，禁止shard做allocate

curl -XPUT localhost:9200/_cluster/settings -d '{

                "persistent" : { 

                    "cluster.routing.allocation.enable" : "none" 

                } 

        }'

3、执行synced操作 

curl -XPOST 'localhost:9200/_flush/synced?pretty'

4、重启es服务

 

5、等级集群变成yellow后开启allocate，允许shard做allocate

curl -XPUT localhost:9200/_cluster/settings -d '{

                "persistent" : { 

                    "cluster.routing.allocation.enable" : "all"

                }

        }'

调整集群恢复时的带宽，-1是指无限制 

curl -XPUT localhost:9200/_cluster/settings -d ' 

{

"transient" : { 

"indices.recovery.max_bytes_per_sec" : "-1" 

} 

} 

'  

调整集群恢复时的单机并发度，最好是和磁盘块数一致 

curl -XPUT localhost:9200/_cluster/settings -d ' 

{ 

"transient" : {

"cluster.routing.allocation.node_concurrent_recoveries" : "24"

} 

} 

'

调整集群恢复时单个shard中同时恢复的小文件的个数

curl -XPUT localhost:9200/_cluster/settings -d '

{

"transient" : {

"indices.recovery.concurrent_small_file_streams" : "24"

}

}

curl -XPUT localhost:9200/_cluster/settings -d '

{

"transient" : { 

"indices.recovery.concurrent_streams" : "24"

}

}

6、等待recovery完成，集群变成green 

 

7、恢复完成后： 

curl -XPUT bapc007:9200/_cluster/settings -d '

{

"transient" : {

"indices.recovery.max_bytes_per_sec" : "100m"

}

}

'  

curl -XPUT bapc007:9200/_cluster/settings -d '

{

"transient" : {

"cluster.routing.allocation.node_concurrent_recoveries" : "12"

} 

}

'

curl -XPUT localhost:9200/_cluster/settings -d '

{

"transient" : {

"indices.recovery.concurrent_small_file_streams" : "6" 

}

}

curl -XPUT localhost:9200/_cluster/settings -d '

{ 

"transient" : {

"indices.recovery.concurrent_streams" : "6" 

}

}