[
https://issues.apache.org/jira/browse/HIVE-25943?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17490008#comment-17490008
]
Janos Kovacs commented on HIVE-25943:
-------------------------------------
If this gets implemented, it would be great to have a retry threshold - time
based - for such abandoned cleaner requests, similar to how initiator ignores
and reprocesses such failed attempts
(compactor.initiator.failed.compacts.threshold &
compactor.initiator.failed.retry.time)
> Introduce compaction cleaner failed attempts threshold
> ------------------------------------------------------
>
> Key: HIVE-25943
> URL: https://issues.apache.org/jira/browse/HIVE-25943
> Project: Hive
> Issue Type: Improvement
> Components: Hive
> Reporter: László Végh
> Assignee: László Végh
> Priority: Major
>
> If the cleaner fails for some reason, the compaction entity status remains in
> "ready for cleaning", therefore the cleaner will pick up this entity
> resulting in an endless try. The number of failed cleaning attempts should be
> counted and if they reach a certain threshold the cleaner must skip all the
> cleaning attempts on that compaction entity.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
