Ceph » CephFS

this crash can be reproduced by 2 steps

1 ceph daemon mds.a scrub_path <dir> recursive
2 ceph daemon mds.a scrub_path <dir> force

crash stack:
2018-01-18 22:52:44.662 7efc3379f700 20 mds.0.cache.ino(0x1) scrub_finished
2018-01-18 22:52:44.662 7efc3379f700 1 mds.0.cache.ino(0x1) 9 != 0
2018-01-18 22:52:44.706 7efc3379f700 -1 /mnt/MyGitPrj/ceph/src/mds/CInode.cc: In function 'void CInode::scrub_finished(MDSInternalContextBase**)' thread 7efc3379f700 time 2018-01-18 22:52:44.668927
/mnt/MyGitPrj/ceph/src/mds/CInode.cc: 4530: FAILED assert(i>second.last_scrub_version == i->second.scrub_start_version)

ceph version 13.0.0-4794-ga3e9911 (a3e9911b841f553517c8f514f6a130a5b5df9a75) mimic (dev)
 1: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x137) [0x7efc4178c350]
 2: (CInode::scrub_finished(MDSInternalContextBase**)+0x3ec) [0x5589393fb99c]
 3: (ScrubStack::_validate_inode_done(CInode*, int, CInode::validated_data const&)+0x6af) [0x55893942fd63]
 4: (C_InodeValidated::finish(int)+0x35) [0x558939431063]
 5: (Context::complete(int)+0x27) [0x558939089cb3]
 6: (MDSInternalContextBase::complete(int)+0x16f) [0x5589394712f3]
 7: (()+0xa69e3b) [0x5589393f8e3b]
 8: (Continuation::continue_function(int, int)+0xc5) [0x5589393ff75d]
 9: (Continuation::Callback::finish(int)+0x2b) [0x5589393ff299]
 10: (Context::complete(int)+0x27) [0x558939089cb3]
 11: (MDSInternalContextWrapper::finish(int)+0x33) [0x558939471353]
 12: (Context::complete(int)+0x27) [0x558939089cb3]
 13: (MDSInternalContextBase::complete(int)+0x16f) [0x5589394712f3]
 14: (CInode::_fetched(ceph::buffer::list&, ceph::buffer::list&, Context*)+0x8f9) [0x5589393df119]
 15: (C_IO_Inode_Fetched::finish(int)+0x40) [0x5589393fff3a]
 16: (Context::complete(int)+0x27) [0x558939089cb3]