cat recovery

The recovery command is a view of index shard recoveries, both on-going and previously completed. It is a more compact view of the JSON recovery API.

A recovery event occurs anytime an index shard moves to a different node in the cluster. This can happen during a snapshot recovery, a change in replication level, node failure, or on node startup. This last type is called a local store recovery and is the normal way for shards to be loaded from disk when a node starts up.

As an example, here is what the recovery state of a cluster may look like when there are no shards in transit from one node to another:

> curl -XGET 'localhost:9200/_cat/recovery?v'
index shard time type  stage source_host  source_node target_host target_node repository snapshot files files_percent bytes bytes_percent
 total_files total_bytes translog translog_percent total_translog
index 0     87ms store done  127.0.0.1        I8hydUG      127.0.0.1        I8hydUG      n/a        n/a      0     0.0%          0     0.0%          0           0           0        100.0%           0
index 1     97ms store done  127.0.0.1        I8hydUG      127.0.0.1        I8hydUG      n/a        n/a      0     0.0%          0     0.0%          0           0           0        100.0%           0
index 2     93ms store done  127.0.0.1        I8hydUG      127.0.0.1        I8hydUG      n/a        n/a      0     0.0%          0     0.0%          0           0           0        100.0%           0
index 3     90ms store done  127.0.0.1        I8hydUG      127.0.0.1        I8hydUG      n/a        n/a      0     0.0%          0     0.0%          0           0           0        100.0%           0
index 4     9ms  store done  127.0.0.1        I8hydUG      127.0.0.1        I8hydUG      n/a        n/a      0     0.0%          0     0.0%          0           0           0        100.0%           0

In the above case, the source and target nodes are the same because the recovery type was store, i.e. they were read from local storage on node start.

Now let’s see what a live recovery looks like. By increasing the replica count of our index and bringing another node online to host the replicas, we can see what a live shard recovery looks like.

> curl -XPUT 'localhost:9200/wiki/_settings' -d'{"number_of_replicas":1}'
{"acknowledged":true}

> curl -XGET 'localhost:9200/_cat/recovery?v&h=i,s,t,ty,st,shost,thost,f,fp,b,bp'
i     s t      ty      st    shost  thost  f     fp      b        bp
wiki  0 1252ms store   done  hostA  hostA  4     100.0%  23638870 100.0%
wiki  0 1672ms replica index hostA  hostB  4     75.0%   23638870 48.8%
wiki  1 1698ms replica index hostA  hostB  4     75.0%   23348540 49.4%
wiki  1 4812ms store   done  hostA  hostA  33    100.0%  24501912 100.0%
wiki  2 1689ms replica index hostA  hostB  4     75.0%   28681851 40.2%
wiki  2 5317ms store   done  hostA  hostA  36    100.0%  30267222 100.0%

We can see in the above listing that our 3 initial shards are in various stages of being replicated from one node to another. Notice that the recovery type is shown as replica. The files and bytes copied are real-time measurements.

Finally, let’s see what a snapshot recovery looks like. Assuming I have previously made a backup of my index, I can restore it using the snapshot and restore API.

> curl -XPOST 'localhost:9200/_snapshot/imdb/snapshot_2/_restore'
{"acknowledged":true}
> curl -XGET 'localhost:9200/_cat/recovery?v&h=i,s,t,ty,st,rep,snap,f,fp,b,bp'
i     s t      ty       st    rep        snap     f     fp      b     bp
imdb  0 1978ms snapshot done  imdb       snap_1   79    8.0%    12086 9.0%
imdb  1 2790ms snapshot index imdb       snap_1   88    7.7%    11025 8.1%
imdb  2 2790ms snapshot index imdb       snap_1   85    0.0%    12072 0.0%
imdb  3 2796ms snapshot index imdb       snap_1   85    2.4%    12048 7.2%
imdb  4  819ms snapshot init  imdb       snap_1   0     0.0%    0     0.0%