I suspect this is the recent deferred_aggressive deadlock; the fix merged last week. Please try the latest master branch and see if you can reproduce. https://github.com/ceph/ceph/pull/16051 Thanks! sage On Mon, 10 Jul 2017, Wangwenfeng wrote: > > Hi, Sage > I setup a Ceph cluster of Luminous 12.0.3, it’s osd use bluestore and I create a cephfs, it’s metadata using replicated and data pool using erasure. > pool 1 'metadata' replicated size 2 min_size 1 crush_ruleset 1 object_hash rjenkins pg_num 1024 pgp_num 1024 last_change 52 flags hashpspool stripe_width 0 > pool 2 'EC_2_1_8' erasure size 3 min_size 2 crush_ruleset 2 object_hash rjenkins pg_num 1024 pgp_num 1024 last_change 56 flags hashpspool,ec_overwrites stripe_width 8192 expected_num_objects 27000000 > When run fio to test the cluster, which command is > fio --numjobs=16 --iodepth=16 --ioengine=libaio --runtime=600 --direct=1 --group_reporting --rw=randwrite --bs=4k --name=aa --filename=/ec/1.txt --size=500G > > for a while time, some osds is reported down, which also in 12.1.0 and master. > > I have a question about follow code > void BlueStore::_deferred_queue(TransContext *txc) { > dout(20) << __func__ << " txc " << txc << " osr " << txc->osr << dendl; > std::lock_guard<std::mutex> l(deferred_lock); > ………………………… > if (deferred_aggressive && > !txc->osr->deferred_running) { > _deferred_submit(txc->osr.get()); > } > } > > If I add '!' to deferred_aggressive, the osd will not down. Would you help to point this modify is right or not? > if (!deferred_aggressive && > !txc->osr->deferred_running) { > _deferred_submit(txc->osr.get()); > } > > > ------------------------------------------------------------------------------------------------------------------------------------- > 本邮件及其附件含有新华三技术有限公司的保密信息，仅限于发送给上面地址中列出 > 的个人或群组。禁止任何其他人以任何形式使用（包括但不限于全部或部分地泄露、复制、 > 或散发）本邮件中的信息。如果您错收了本邮件，请您立即电话或邮件通知发件人并删除本 > 邮件！ > This e-mail and its attachments contain confidential information from New H3C, which is > intended only for the person or entity whose address is listed above. Any use of the > information contained herein in any way (including, but not limited to, total or partial > disclosure, reproduction, or dissemination) by persons other than the intended > recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender > by phone or email immediately and delete it! >
