I've got a single shard index of 75GB taking 40% of my physical disk space (according to
df) which I needed to split into two shards. After the split, the disk space taken by the new index is around the double of that (175GB according to Kibana/Index Management) and slowly going down (which is expected AFAIU). Both indexes now take only 43% of my disk space (according to
df) and slowly going up: this also seems to be expected since, IIUC, the hard links created during the split are progressively being replaced by copied data. Please, correct me if I am wrong.
My question is: when is it safe to remove the original index from my cluster? I expect that I have to wait for the whole splitting process to be done. Is there a command that I could use to monitor the disk space operations that are taking place? I've tried to monitor using
GET _cat/recovery?v that returns the following:
index shard time type stage source_host source_node target_host target_node repository snapshot files files_recovered files_percent files_total bytes bytes_recovered bytes_percent bytes_total translog_ops translog_ops_recovered translog_ops_percent original_index 0 48.8s existing_store done n/a n/a 172.18.0.2 es-primary n/a n/a 0 0 100.0% 356 0 0 100.0% 77571827169 0 0 100.0% new_index 0 2.6m local_shards done n/a n/a 172.18.0.2 es-primary n/a n/a 0 0 100.0% 355 0 0 100.0% 77571822082 0 0 100.0% new_index 1 2.6m local_shards done n/a n/a 172.18.0.2 es-primary n/a n/a 0 0 100.0% 355 0 0 100.0% 77571822082 0 0 100.0%
As far as I understand, there's no recovery to be done or that would be in progress, unless I am looking at wrong places. Could someone guide me toward the right monitoring process and find the right time to delete original_index like I am five?
Thanks a lot!