Add note about ILM and Snapshots #36023
Conversation
This commit documents a workaround for situations where an index may be deleted by ILM immediately afterit is restored from a snapshot.
|
Pinging @elastic/es-core-features |
| restored index to be deleted by Index Lifecycle Management immediately after it | ||
| is restored. To prevent this, before restoring the snapshot, | ||
| <<start-stop-ilm,lifecycle policy execution can be paused>> to allow the policy | ||
| to be removed. |
There was a problem hiding this comment.
I think we should re-word this to describe that in some case you may not want the lifecycle to continue to execute after the restore and in those case you should use the process described. This is for two reasons: 1) This is not a bug but the above wording may come across sounding like a bug to some users and 2) there are other reasons that having a delete action that you might not want the index to continue on its lifecycle, such as if it would allocate to a node attribute that no longer exists or if the policy no longer exists.
There was a problem hiding this comment.
Thanks - I've rewritten it to be clear that this is the intended behavior and extended it to other examples besides index deletion.
This commit documents how Index Lifecycle Management interacts with snapshot/restore, and documents a workaround for situations in which ILM should not immediately resume managing an index after it is restored.
This commit documents a workaround for situations where an index
may be deleted by ILM immediately after it is restored from a snapshot.
Documents a workaround for #35561