Handle index deletion while querying in ES|QL #137702
Conversation
dnhatn
force-pushed
the
query-while-delete-index
branch
from
1ef957c to
ffcec0e
Compare
|
Hi @dnhatn, I've created a changelog YAML for you. |
|
Hi @dnhatn, I've updated the changelog YAML for you. |
elasticsearchmachine
added
the
Team:Analytics
|
Pinging @elastic/es-analytical-engine (Team:Analytics) |
| } | ||
| handler.messageReceived(request, channel, task); | ||
| }); | ||
| } |
| private void expectNoFailure(String message, Exception e) { | ||
| LOGGER.error(message, e); | ||
| assert false : new AssertionError(message, e); | ||
| } |
There was a problem hiding this comment.
I believe the change to this file says "we expect no failures handling the response and if there is one log something big and loud and, if we're testing, crash loudly so we see if and fix it right away. If we're in production just be loud."
There was a problem hiding this comment.
Yes, that's correct. In production, if a request enters this state, we will fail it fast and log the error.
| .toList(); | ||
| } catch (Exception ignored) { | ||
| continue; | ||
| } |
There was a problem hiding this comment.
Should we continue on any exception here or just the one we get from the index being deleted. Is that just IndexNotFoundException?
Would you kindly add a comment around continue; saying "this is a shard being deleted after we started the request"?
3 participants
If the index is deleted while querying, ES|QL might try to get the index routing from the latest cluster state. This could throw an index-not-found exception, which is not handled properly, causing the user to never receive a response, even though execution has completed. This also results in a dangling task in the task manager.
Relates #126653
Closes #135863