Bug #48819: fsck error: found stray (per-pg) omap data on omap_head - bluestore - Ceph

Actions

Copy link

Bug #48819

open

fsck error: found stray (per-pg) omap data on omap_head

Added by Kefu Chai over 3 years ago. Updated almost 2 years ago.

Status:

New

Priority:

Normal

Assignee:

Target version:

% Done:

Source:

Tags:

Backport:

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

ceph-qa-suite:

Pull request ID:

Crash signature (v1):

Crash signature (v2):

Description

/a/kchai-2021-01-10_13:20:22-rados-master-distro-basic-smithi/

Related issues 1 (0 open — 1 closed)

Actions

Copy link

Updated by Josh Durgin over 3 years ago

This is happening due to an exception running ceph-objectstore-tool:

2021-01-10T13:59:28.286 ERROR:tasks.thrashosds.thrasher:exception:
Traceback (most recent call last):
  File "/home/teuthworker/src/github.com_ceph_ceph_master/qa/tasks/ceph_manager.py", line 1069, in do_thrash
    self._do_thrash()
  File "/home/teuthworker/src/github.com_ceph_ceph_master/qa/tasks/ceph_manager.py", line 115, in wrapper
    return func(self)
  File "/home/teuthworker/src/github.com_ceph_ceph_master/qa/tasks/ceph_manager.py", line 1201, in _do_thrash
    self.choose_action()()
  File "/home/teuthworker/src/github.com_ceph_ceph_master/qa/tasks/ceph_manager.py", line 308, in kill_osd
    format(ret=proc.exitstatus))
Exception: ceph-objectstore-tool: exp list-pgs failure with status 1

This halts execution of the thrasher thread, since nothing handles the exception, which leaves the cluster with at least one osd down. Thus, adding exception handling for these ceph-object-store calls would fix the problem.

In terms of why ceph-objectstore-tool falied, it may be due to an osd not being completely stopped when trying to run it.

Actions

Copy link

Updated by Josh Durgin over 3 years ago

Project changed from RADOS to bluestore

Cause appears to be a bluestore issue - manually running the objectstore command with --log-to-stderr we can see it's failing fsck with stray per-pg omap data:

https://pulpito.ceph.com/yuriw-2021-01-15_19:06:33-rados-wip-yuri8-testing-master-2021-01-15-0935-distro-basic-smithi/5789365/

2021-01-15T23:56:00.067+0000 7f77a73d6dc0  1 bluefs fsck
2021-01-15T23:56:00.067+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open walking object keyspace
2021-01-15T23:56:00.069+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open checking shared_blobs
2021-01-15T23:56:00.069+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open checking pool_statfs
2021-01-15T23:56:00.069+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open checking for stray omap data 
2021-01-15T23:56:00.069+0000 7f77a73d6dc0 -1 bluestore(/var/lib/ceph/osd/ceph-4) fsck error: found stray (per-pg) omap data on omap_head 16092768843478336515 0 0
2021-01-15T23:56:00.069+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open checking deferred events
2021-01-15T23:56:00.069+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open checking freelist vs allocated
2021-01-15T23:56:00.141+0000 7f77a73d6dc0  1 bluestore(/var/lib/ceph/osd/ceph-4) _fsck_on_open <<<FINISH>>> with 1 errors, 0 warnings, 0 repaired, 1 remaining in 0.075985 seconds
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4)  allocation stats probe 0: cnt: 0 frags: 0 size: 0
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4)  probe -1: 0,  0, 0
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4)  probe -2: 0,  0, 0
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4)  probe -4: 0,  0, 0
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4)  probe -8: 0,  0, 0
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4)  probe -16: 0,  0, 0
2021-01-15T23:56:00.141+0000 7f7769ffb700  0 bluestore(/var/lib/ceph/osd/ceph-4) ------------
2021-01-15T23:56:00.141+0000 7f77a73d6dc0  4 rocksdb: [db_impl/db_impl.cc:397] Shutdown: canceling all background work
2021-01-15T23:56:00.141+0000 7f77a73d6dc0  4 rocksdb: [db_impl/db_impl.cc:573] Shutdown complete
2021-01-15T23:56:00.141+0000 7f77a73d6dc0  1 bluefs umount
2021-01-15T23:56:00.141+0000 7f77a73d6dc0  1 bdev(0x55dfbbe73bb0 /var/lib/ceph/osd/ceph-4/block) close
2021-01-15T23:56:00.321+0000 7f77a73d6dc0  1 freelist shutdown
2021-01-15T23:56:00.321+0000 7f77a73d6dc0  1 stupidalloc 0x0x55dfbbe7f7e0 shutdown
2021-01-15T23:56:00.321+0000 7f77a73d6dc0  1 bdev(0x55dfbbe3ba00 /var/lib/ceph/osd/ceph-4/block) close
Mount failed with '(5) Input/output error'
2021-01-15T23:56:00.554+0000 7f77a73d6dc0 -1 bluestore(/var/lib/ceph/osd/ceph-4) _mount fsck found 1 errors

Actions

Copy link

Updated by Neha Ojha over 3 years ago

Related to Bug #48810: "mount fsck found 1 errors" in crimson-rados-master added

Actions

Copy link

Updated by Igor Fedotov over 3 years ago

This looks related to my recent PR introducing per-pg omap naming scheme: https://github.com/ceph/ceph/pull/38651

Wondering how I can manually (and locally) run the test case to reproduce the issue?
Or at least some insight on the test case scenario would be highly appreciated.

Actions

Copy link

Updated by Josh Durgin over 3 years ago

Igor, you can log on to ubuntu@smithi092 right now and check out osd.4.

The test isn't doing much special, it's thrashing osds (taking them up/down) and testing monitor recovery. Not doing much I/O other than osdmaps I would think.

Actions

Copy link