目錄

ElasticSearch 維運 - Shrink Shard

Shrink the number of shards

Brief introduction

  1. Move all indices data into the same data node and block writing
  2. Check data moving status util finish
  3. Call _shrink api and assign the new name and settings(included # of shards)
  4. Check the new indices creation status
  5. Set the replica back, and check the new one and old one whether the doc.count is the same or not.
  6. Delete old indices. (Be careful for all deletion, double check the snapshot before the deletion)

Example

If we want to shrink the number of shards from 4 to 1 on log-202105-0503-000005 indices.

health status index                     uuid   pri rep docs.count docs.deleted store.size pri.store.size
green  open   log-202105-0530-000005 xxxx   4   0   14845131            0     28.5gb         28.5gb

Run Step 1.

  • Choose one unbusy data node and fill in index.routing.allocation.require._name"
  • Set replica to 0 (for speed up whole progress)
  • Block write action
警告
Change indices target and node name
PUT /log-202105-0530-000005/_settings
{
  "settings": {
    "index.number_of_replicas": 0,
    "index.routing.allocation.require._name": "X1imzcc",
    "index.blocks.write": true
  }
}

Run Step 2. Check Status

GET _cat/recovery/log-202105-0530-000005?v&detailed&h=s,t,ty,source_node,target_node,files_recovered,bytes_percent&s=bytes_percent:desc

There are all in the same target_node (X1imzcc).

s t    ty   source_node target_node files_recovered bytes_percent
0 2.2m peer bzGs44F     X1imzcc     113             100.0%
1 2.2m peer HWatjUI     X1imzcc     107             100.0%
2 2.2m peer -3AWCw0     X1imzcc     113             100.0%
3 1.9m peer o3pEUt9     X1imzcc     118             100.0%

Most of time you have to block again, if you encounter the error about READ-ONLY issus

PUT /log-202105-0530-000005/_settings
{
  "settings": {
      "index.blocks.write": true                                    
  }
}

Run Step 3. do _shrink action

  • number_of_replica: 0 and refresh_interval: -1 is for speedup action
警告
Change indices target and the new indices name, and how many number of shards do you assign
POST /log-202105-0530-000005/_shrink/log-202105-0530-000005-s
{
  "settings": {
    "index": {
      "number_of_shards": 1,
      "number_of_replicas": 0,
      "refresh_interval": -1
    },
    "index.routing.allocation.require._name": null,
    "index.blocks.write": null
  }
}

Run Step 4. check status

GET /_cat/recovery/log-202105-0530-000005-s/?v&h=index,shard,time
index                       shard time
log-202105-0530-000005-s 0     16.3s
log-202105-0530-000005-s 0     3.2s

Run Step 5. Set replica back

PUT log-202105-0530-000005-s/_settings
{
  "index": {
  "number_of_replicas": 1,
  "refresh_interval": null
  }
}
GET _cat/indices/log-202105-0530-000005*?v&s=index
health status index                    uuid  pri rep docs.count store.size pri.store.size
green  open   log-202105-0530-000005   xxxx   4   0   14845131     28.5gb         28.5gb
yellow open   log-202105-0530-000005-s xxxx   1   1   14845131     28.2gb         28.2gb

Check the number of docs.count and health util all are ready.

Run Step 6. Delete all indices

DELETE log-202105-0530-000005