Project

General

Profile

Actions

Bug #51256

closed

pybind/mgr/volumes: purge queue seems to block operating on cephfs connection required by dispatch thread

Added by Patrick Donnelly almost 3 years ago. Updated over 2 years ago.

Status:
Resolved
Priority:
Immediate
Assignee:
Category:
-
Target version:
% Done:

0%

Source:
other
Tags:
Backport:
pacific
Regression:
No
Severity:
1 - critical
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(FS):
mgr/volumes
Labels (FS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

debug 2021-06-16 13:38:35.555 7f724dcde700  3 client.96349150 read(10, 0x55a35f3ae020, 8192, -1) = 158
...
debug 2021-06-16 13:38:35.555 7f7218edd700 10 client.96349150 path_walk #0x1/volumes/_deleting/path/to/dir
...
debug 2021-06-16 13:38:45.625 7f7218edd700 10 client.96349150  marking I_COMPLETE on 0x1001b524ede.head(faked_ino=0 ref=5 ll_ref=0 cap_refs={} open={} mode=40755 size=0/0 nlink=1 btime=2020-10-18 19:09:08.080929 mtime=2021-06-16 13:38:35.557210 ctime=2021-06-16 13:38:35.557210 caps=pAsLsXsFsx(0=pAsLsXsFsx) parents=0x1001b524ed3.head["end"] 0x55a3607c6680)
...
debug 2021-06-16 13:38:45.626 7f724dcde700 10 client.96349150 lock_fh_pos 0x55a3604c4420
debug 2021-06-16 13:38:45.626 7f724dcde700 10 client.96349150 get_caps 0x1006a9ebb21.head(faked_ino=0 ref=2 ll_ref=0 cap_refs={4=0,1024=0,2048=0,4096=0,8192=0} open={1=1,2=0} mode=100640 size=158/4194304 nlink=1 btime=2021-06-02 21:25:26.961098 mtime=2021-06-16 13:38:32.702209 ctime=2021-06-16 13:38:32.702209 caps=pAsxLsXsxFsxcrwb(0=pAsxLsXsxFsxcrwb) objectset[0x1006a9ebb21 ts 2/0 objects 1 dirty_or_tx 0] parents=0x1006a9ebb1f.head[".meta"] 0x55a3605cbb80) have pAsxLsXsxFsxcrwb need Fr want Fc revoking -

7f724dcde700 is the finisher thread. 7f7218edd700 appears to be one of the purge queue threads. There seems to be a 10 second gap in processing by the finisher thread executing a subvolume API operation. The client instance "client.96349150" is used by both threads.


Related issues 2 (0 open2 closed)

Related to CephFS - Bug #51271: mgr/volumes: use a dedicated libcephfs handle for subvolume API callsResolvedVenky Shankar

Actions
Copied to CephFS - Backport #51411: pacific: pybind/mgr/volumes: purge queue seems to block operating on cephfs connection required by dispatch threadResolvedKotresh Hiremath RavishankarActions
Actions

Also available in: Atom PDF