es集群快速恢复（优化方案）

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"

 

}

 

}