Actions
Bug #21070
closedMDS: MDS is laggy or crashed When deleting a large number of files
% Done:
0%
Source:
Tags:
Backport:
luminous
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
fs
Component(FS):
Labels (FS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
We plan to use mdtest to create 100w level of the file, in the ceph-fuse mount the directory, the command is as follows:
./mdtest -d / cephfs / test / -z 1 -w 10k 1 -T -b 10 -L -n 1000000 -F -C
The creation process is normal, but when we delete the file, the rm -rf command is used,the error happened, as follows:
services: mon: 2 daemons, quorum 192.9.9.221,192.9.9.223 mgr: 192.9.9.221(active), standbys: 192.9.9.223 mds: 1/1/1 up {0=192.9.9.223=up:active(laggy or crashed)} osd: 19 osds: 19 up, 19 in
The mds log has the following error:
2017-08-22 15:29:11.017122 7ffb752f8700 1 mds.0.28 recovery_done -- successful recovery! 2017-08-22 15:29:11.017247 7ffb752f8700 1 mds.0.28 active_start 2017-08-22 15:29:11.018963 7ffb752f8700 1 mds.0.28 cluster recovered. 2017-08-22 15:29:11.023689 7ffb6f2ec700 -1 *** Caught signal (Segmentation fault) ** in thread 7ffb6f2ec700 thread_name:mds_rank_progr ceph version 12.1.0.3 (48ff484a1b75334f6a08c5b3dbdbb0abfa1cb2cf) luminous (dev) 1: (()+0x58592f) [0x7ffb7e1d792f] 2: (()+0xf130) [0x7ffb7bd25130] 3: (Server::handle_client_readdir(boost::intrusive_ptr<MDRequestImpl>&)+0xbb9) [0x7ffb7df54989] 4: (Server::dispatch_client_request(boost::intrusive_ptr<MDRequestImpl>&)+0x9b1) [0x7ffb7df85a51] 5: (MDSInternalContextBase::complete(int)+0x1eb) [0x7ffb7e162dcb] 6: (MDSRank::_advance_queues()+0x4a5) [0x7ffb7df0d665] 7: (MDSRank::ProgressThread::entry()+0x4a) [0x7ffb7df0daca] 8: (()+0x7df3) [0x7ffb7bd1ddf3] 9: (clone()+0x6d) [0x7ffb7ae033ed] NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.
Description:
12.1.0.3 comes from version 12.1.0
Files
Actions