https://tracker.ceph.com/https://tracker.ceph.com/favicon.ico2016-05-06T07:08:16ZCeph Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=701662016-05-06T07:08:16ZHaomai Wanghaomaiwang@gmail.com
<ul></ul><p>The lock time depends on the ObjectStore process time from 1s-5s even longer. It's easy to cause IO shake when exists non-fast dispatch message in.</p> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=701682016-05-06T07:48:14ZHaomai Wanghaomaiwang@gmail.com
<ul></ul><p>It looks like async messenger need a extra DispatchQueue to handle nonfast message which may block thread totally.</p> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=701692016-05-06T07:48:41ZHaomai Wanghaomaiwang@gmail.com
<ul><li><strong>Subject</strong> changed from <i>Messenger thread long time lock hold risk</i> to <i>msgr/async: Messenger thread long time lock hold risk</i></li><li><strong>Category</strong> set to <i>msgr</i></li></ul> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=701952016-05-06T19:01:11ZGreg Farnumgfarnum@redhat.com
<ul></ul><p>Is this something that newly blocks for a long time?</p>
<p>Or is the problem that AsyncMessenger doesn't have per-connection threads and so all the other message processing gets blocked up? (And so this has been a problem for a long time, but SimpleMessenger masked it.)</p> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=702032016-05-07T02:34:58ZHaomai Wanghaomaiwang@gmail.com
<ul></ul><p>yes, async messenger always easy to expose problem than simple. actually this pr(<a class="external" href="https://github.com/ceph/ceph/pull/8808">https://github.com/ceph/ceph/pull/8808</a>) help a lot. If FileStore is busy in sync, a nonfast message like pg_log(from create pool action) will hold osd_lock and stuck into filestore condition wait. After we have OSDPing message fast dispatch, it could avoid osd down because of heartbeat timeout while ping message in the queue.</p> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=729622016-06-17T14:26:16ZSage Weilsage@newdream.net
<ul><li><strong>Status</strong> changed from <i>New</i> to <i>Pending Backport</i></li><li><strong>Backport</strong> set to <i>jewel</i></li></ul><p>let's take our time backporting this... it should bake in master for a while first!</p> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=729862016-06-17T18:01:29ZNathan Cutlerncutler@suse.cz
<ul><li><strong>Copied to</strong> <i><a class="issue tracker-9 status-3 priority-4 priority-default closed" href="/issues/16377">Backport #16377</a>: jewel: msgr/async: Messenger thread long time lock hold risk</i> added</li></ul> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=797212016-10-13T15:51:25ZLoïc Dacharyloic@dachary.org
<ul><li><strong>Status</strong> changed from <i>Pending Backport</i> to <i>Resolved</i></li></ul> Messengers - Bug #15758: msgr/async: Messenger thread long time lock hold riskhttps://tracker.ceph.com/issues/15758?journal_id=1317662019-03-12T23:12:32ZGreg Farnumgfarnum@redhat.com
<ul><li><strong>Project</strong> changed from <i>Ceph</i> to <i>Messengers</i></li><li><strong>Category</strong> deleted (<del><i>msgr</i></del>)</li></ul>