[
https://issues.apache.org/jira/browse/HDDS-16845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated HDDS-16845:
----------------------------------
Labels: pull-request-available (was: )
> Recon delta updates to om.db are never synced and ignore configured write
> options
> ---------------------------------------------------------------------------------
>
> Key: HDDS-16845
> URL: https://issues.apache.org/jira/browse/HDDS-16845
> Project: Apache Ozone
> Issue Type: Task
> Components: Ozone Recon
> Reporter: Devesh Kumar Singh
> Assignee: Devesh Kumar Singh
> Priority: Major
> Labels: pull-request-available
>
> Recon applies OM delta updates to its local om.db using new RocksDB write
> options that disable sync and bypass the configured store write options.
> Therefore, hadoop.hdds.db.rocksdb.writeoption.sync=true has no effect.
> After a host crash, om.db and the derived data in recon.db can recover to
> different sequence numbers. Recon may then replay already-processed events or
> skip missing events, causing permanent over-counting or under-counting.
> The fix should persist each Recon task's processing cursor atomically with
> its derived data in recon.db, allowing startup recovery to detect
> inconsistencies. At minimum, delta writes to om.db must honor the configured
> RocksDB write options.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
