[
https://issues.apache.org/jira/browse/HDDS-15036?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Devesh Kumar Singh resolved HDDS-15036.
---------------------------------------
Resolution: Duplicate
> Retry replication/import on another volume when destination container
> directory already exists
> ----------------------------------------------------------------------------------------------
>
> Key: HDDS-15036
> URL: https://issues.apache.org/jira/browse/HDDS-15036
> Project: Apache Ozone
> Issue Type: Task
> Components: Ozone Datanode
> Affects Versions: 2.1.0
> Reporter: Devesh Kumar Singh
> Assignee: Devesh Kumar Singh
> Priority: Major
>
> *Description:*
> During DN-side replication/import, the current path selection validates
> only the chosen target path. If a stale container directory already exists on
> that selected volume, the import/create path can fail with
> CONTAINER_ALREADY_EXISTS even when another eligible volume is available on
> the same DN.
> *Current behavior:*
> - a target volume is chosen once
> - the container create/import path checks only the selected target metadata
> path
> - if that path already exists, the operation fails
> - the flow does not appear to retry on another eligible volume
> *Why this is a problem:*
> A stale undeleted directory on one volume should not necessarily convert a
> per-volume collision into a full DN-level replication/import failure. This is
> especially relevant when stale directories may remain after earlier failed
> delete or duplicate-directory scenarios.
> *Proposed fix:*
> - treat destination-path collision as a recoverable per-volume failure in
> DN replication/import flows
> - when the selected volume already contains the target container directory,
> retry with another eligible volume if available
> - keep the check scoped to the selected volume/path rather than introducing
> a DN-wide scan unless later proven necessary
> *Acceptance criteria:*
> - replication/import succeeds on another eligible volume when one selected
> volume has a stale destination directory
> - if no alternate volume exists, the failure remains explicit
> - add tests for stale destination directory on one volume with successful
> retry to another
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]