All of lore.kernel.org
 help / color / mirror / Atom feed
* rgw/multisite: unlock failure problem
@ 2016-11-14  6:40 Tianshan Qu
  2016-11-14 21:26 ` Yehuda Sadeh-Weinraub
  0 siblings, 1 reply; 3+ messages in thread
From: Tianshan Qu @ 2016-11-14  6:40 UTC (permalink / raw)
  To: ceph-devel

Hi:

There is a problem with unlock multiste sync log.
The situation is set rgw_num_rados_handles > 1 and
rgw_num_async_rados_threads default 32. since multisite use async
thread pool, so lock and unlock operation may get different threads,
which may direct to different rados handles, and lock operation use
client.xxxx to identify owner, different rados handles means different
client, leads to unlock fail, and next lock will return locked, rgw
log will show many -16 error.
Is there a good way to solve this?

^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: rgw/multisite: unlock failure problem
  2016-11-14  6:40 rgw/multisite: unlock failure problem Tianshan Qu
@ 2016-11-14 21:26 ` Yehuda Sadeh-Weinraub
  2016-11-15  8:30   ` Tianshan Qu
  0 siblings, 1 reply; 3+ messages in thread
From: Yehuda Sadeh-Weinraub @ 2016-11-14 21:26 UTC (permalink / raw)
  To: Tianshan Qu; +Cc: ceph-devel

On Sun, Nov 13, 2016 at 10:40 PM, Tianshan Qu <qutianshan@gmail.com> wrote:
> Hi:
>
> There is a problem with unlock multiste sync log.
> The situation is set rgw_num_rados_handles > 1 and
> rgw_num_async_rados_threads default 32. since multisite use async
> thread pool, so lock and unlock operation may get different threads,
> which may direct to different rados handles, and lock operation use
> client.xxxx to identify owner, different rados handles means different
> client, leads to unlock fail, and next lock will return locked, rgw
> log will show many -16 error.
> Is there a good way to solve this?

Can you open a ceph tracker issue for that?

Thanks,
Yehuda

^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: rgw/multisite: unlock failure problem
  2016-11-14 21:26 ` Yehuda Sadeh-Weinraub
@ 2016-11-15  8:30   ` Tianshan Qu
  0 siblings, 0 replies; 3+ messages in thread
From: Tianshan Qu @ 2016-11-15  8:30 UTC (permalink / raw)
  To: Yehuda Sadeh-Weinraub; +Cc: ceph-devel

http://tracker.ceph.com/issues/17907

2016-11-15 5:26 GMT+08:00 Yehuda Sadeh-Weinraub <yehuda@redhat.com>:
> On Sun, Nov 13, 2016 at 10:40 PM, Tianshan Qu <qutianshan@gmail.com> wrote:
>> Hi:
>>
>> There is a problem with unlock multiste sync log.
>> The situation is set rgw_num_rados_handles > 1 and
>> rgw_num_async_rados_threads default 32. since multisite use async
>> thread pool, so lock and unlock operation may get different threads,
>> which may direct to different rados handles, and lock operation use
>> client.xxxx to identify owner, different rados handles means different
>> client, leads to unlock fail, and next lock will return locked, rgw
>> log will show many -16 error.
>> Is there a good way to solve this?
>
> Can you open a ceph tracker issue for that?
>
> Thanks,
> Yehuda

^ permalink raw reply	[flat|nested] 3+ messages in thread
end of thread, other threads:[~2016-11-15  8:31 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-11-14  6:40 rgw/multisite: unlock failure problem Tianshan Qu
2016-11-14 21:26 ` Yehuda Sadeh-Weinraub
2016-11-15  8:30   ` Tianshan Qu

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.