[
https://issues.apache.org/jira/browse/HDDS-8191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17731812#comment-17731812
]
Ethan Rose commented on HDDS-8191:
----------------------------------
This can be handled by running an on-demand volume scan after the container
scan has failed, but before marking it unhealthy. If the volume scan fails, we
should not mark the container unhealthy and fail the volume instead. After
HDDS-8448, failing the volume will remove the containers from the datanode, not
mark them unhealthy.
> Avoid false positive failures in scanners
> -----------------------------------------
>
> Key: HDDS-8191
> URL: https://issues.apache.org/jira/browse/HDDS-8191
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Ethan Rose
> Priority: Major
>
> Sometimes volume or container scan may fail due to external issues in the
> file system or operation system that do not affect the content of the
> containers or volumes and are caused by simple admin mistakes. This jira
> proposes having the following failures be logged by the scanners but not
> cause their contents to be marked unhealthy:
> * Too many open files
> ** Can be fixed by the admin adjusting the file descriptor limit for the node.
> * Permission denied for reading or writing.
> ** Can be fixed by the admin fixing filesystem permissions.
> * No space left on device
> ** Should be prevented by configuring hdds.datanode.dir.du.reserved.percent
> to leave some free space on the volume.
> ** Failing full volumes could put an already full cluster in a worse state by
> triggering replication.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@ozone.apache.org
For additional commands, e-mail: issues-h...@ozone.apache.org
