OK here is one thing you could do if you are able to recreate this hang
easily on your setup.
The next time you find heal-info hanging, could you do the following:

# gluster volume statedump <VOL>

and share the statedump logs? You will find them at /var/run/gluster/

-Krutika

On Thu, Apr 14, 2016 at 3:32 PM, Krutika Dhananjay <[email protected]>
wrote:

> Would you happen to know what those 6 entries that need heal correspond
> to? Assuming heal-info reported the status at least once without hanging.
> Also, could you share the contents of glfsheal-datastore.log, specifically
> from the machine where you ran the `heal info` command?
>
> -Krutika
>
> On Wed, Apr 13, 2016 at 8:13 PM, Lindsay Mathieson <
> [email protected]> wrote:
>
>> On 14/04/2016 12:19 AM, Krutika Dhananjay wrote:
>>
>>> Hmm what version of gluster was the hang seen on?
>>>
>>
>>
>> Ah yes, sorry - 3.7.9
>>
>> The heal was triggered by a "killall glusterfsd" on node vna as a test, I
>> then restarted the glusterd service to bring it back up.
>>
>> Eventually I shutdown the VM's and stopped all gluster servcies, then
>> brought them back up. Eventually the last 4 shards healed, but it took
>> around an hour.
>>
>> Prior to that I had done rolling updates and restarts across all three
>> node, heals between reboots completed reasonably quickly with no i/o
>> freezes.
>>
>> I off to bed now I'm afraid. Rather late here :)
>>
>> --
>> Lindsay Mathieson
>>
>>
>
_______________________________________________
Gluster-users mailing list
[email protected]
http://www.gluster.org/mailman/listinfo/gluster-users

Reply via email to