You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by "Attila Doroszlai (Jira)" <ji...@apache.org> on 2024/09/17 10:12:00 UTC
[jira] [Resolved] (HDDS-11453) OmSnapshotPurge should be in a different ozone manager double buffer batch
[ https://issues.apache.org/jira/browse/HDDS-11453?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Attila Doroszlai resolved HDDS-11453.
-------------------------------------
Fix Version/s: 1.5.0
Resolution: Fixed
> OmSnapshotPurge should be in a different ozone manager double buffer batch
> --------------------------------------------------------------------------
>
> Key: HDDS-11453
> URL: https://issues.apache.org/jira/browse/HDDS-11453
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Swaminathan Balachandran
> Assignee: Swaminathan Balachandran
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.5.0
>
>
> Since purgeSnapshot transaction on a standalone basis is an idempotent operation. Once the directory gets deleted the previous transactions that have been performed on the snapshotted rocksdb would start failing on replay since those transactions have not been committed but the directory could have been partially deleted/ fully deleted. This could also lead to inconsistencies in the DB reads from the purged rocksdb if operations are not performed consciously.
> Creating a separate rocksdb batch in ozone manager double buffer will ensure all transactions before have been flushed to the disk before purging the rocksdb, ensuring the previous ratis transactions will never get replayed.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org