On 04/29/2015 03:21 AM, John Gardeniers wrote:> Hi All,
>
> We have a replica 2 Gluster set-up and I noticed that there was one
> file showing as heal-failed:
>
> # gluster volume heal gluster-rhev info heal-failed
> Gathering list of heal failed entries on volume gluster-rhev has been
> successful
>
> Brick jupiter.om.net:/gluster_brick_1
> Number of entries: 1
> at path on brick
> -----------------------------------
> 2015-03-22 21:29:37
>
/44d30b24-1ed7-48a0-b905-818dc0a006a2/images/acce7c96-fcd1-483a-9745-b7dc9a19f5b0/8d192146-2b5c-4195-b535-12e06231e0d6
>
> Brick rigel:/gluster_brick_1
> Number of entries: 0
>
> The file in question is a VM image. It has since been removed via the
> client (RHEV). I've verified that the file no longer exists on either
> replica. However, an hour later I still get the same heal-failed report.
>
> How can I clean up this extraneous entry? Taking Gluster off-line is
> not an option at this time.
>
It is safe to ignore the stale entries. If you really want to get rid of
them, you can do a volume start force, which will restart the self-heal
daemon (glustershd) and clear its buffers. In recent releases, 'info
healed' and info heal-failed' commands have been removed as they
don't
give much insight. All you need to track is the output of 'heal info' If
a file needs healing, it should appear in heal info's output.
HTH,
Ravi
> Cheers,
> John
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-users