Lockups during hammer cleanup

Matthew Dillon dillon at backplane.com
Sat Jul 23 17:50:28 PDT 2016


Yah, we're still seeing lockups as well.  Still tracking it down.  At least
we can reproduce them (just not very quickly).  I'll keep people posted.

-Matt

On Sat, Jul 23, 2016 at 7:27 AM, Tim Darby <t+dfbsd at timdarby.net> wrote:

> I'm still seeing lock ups on reblock and recopy after last night's changes.
>
> Tim
>
> On Fri, Jul 22, 2016 at 12:22 PM, Matthew Dillon <dillon at backplane.com>
> wrote:
>
>> I've been able to reproduce the buildworld lockup, this is probably the
>> same thing.  I hope to track it down today.
>>
>> -Matt
>>
>> On Fri, Jul 22, 2016 at 11:53 AM, Tim Darby <t+dfbsd at timdarby.net> wrote:
>>
>>> I'm seeing this on two different machines for the same hammer filesystem
>>> (one machine is the master, the other is the slave for that fs):
>>>
>>> hammer -v cleanup /volumes/BACKUP2/pfs/media
>>> cleanup /volumes/BACKUP2/pfs/media - handle PFS #1 using
>>> /var/hammer/volumes/BACKUP2/pfs/media
>>>            snapshots - skip
>>>                prune - skip
>>>            rebalance - skip
>>>              reblock - run
>>>     hammer -c /var/hammer/volumes/BACKUP2/pfs/media/.reblock-1.cycle -t
>>> 300 reblock-btree /volumes/BACKUP2/pfs/media
>>>
>>> <machine locks up here, nothing on the console>
>>>
>>> I was able to get it to complete eventually by reducing the time for
>>> reblock and recopy to 30 seconds. If it goes longer than that, it locks.
>>>
>>> Tim
>>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.dragonflybsd.org/pipermail/users/attachments/20160723/46eb102f/attachment-0004.html>


More information about the Users mailing list