Bug #1310
closed
mds crash 'T interval_set<T>::range_start() const [with T = inodeno_t]'
Added by Damien Churchill almost 13 years ago.
Updated over 7 years ago.
Description
I have a reproducible mds crash, I've attached the last 200 lines of the log with debug mds = 20.
Files
End of log:
2011-07-14 09:05:57.848753 7f8f4608a700 mds2.inotable: project_alloc_id 0 to []/[]
./include/interval_set.h: In function 'T interval_set<T>::range_start() const [with T = inodeno_t]', in thread '0x7f8f4608a700'
./include/interval_set.h: 290: FAILED assert(!empty())
ceph version 0.31 (commit:9019c6ce64053ad515a493e912e2e63ba9b8e278)
1: /usr/bin/cmds() [0x49305e]
2: (InoTable::project_alloc_id(inodeno_t)+0x408) [0x636728]
3: (Server::prepare_new_inode(MDRequest*, CDir*, inodeno_t, unsigned int, ceph_file_layout*)+0x9cb) [0x4fba1b]
4: (Server::handle_client_mkdir(MDRequest*)+0x2be) [0x50746e]
5: (Server::dispatch_client_request(MDRequest*)+0x4eb) [0x5183cb]
6: (MDCache::dispatch_request(MDRequest*)+0x41) [0x531571]
7: (C_MDS_RetryRequest::finish(int)+0x11) [0x51d581]
8: (finish_contexts(CephContext*, std::list<Context*, std::allocator<Context*> >&, int)+0xc2) [0x4ba4e2]
9: (SimpleLock::finish_waiters(unsigned long, int)+0xad) [0x5c506d]
10: (Locker::eval_gather(SimpleLock*, bool, bool*, std::list<Context*, std::allocator<Context*> >*)+0x126f) [0x5b34ef]
11: (Locker::handle_file_lock(ScatterLock*, MLock*)+0xa02) [0x5b4232]
12: (Locker::handle_lock(MLock*)+0x1c6) [0x5b5446]
13: (MDS::handle_deferrable_message(Message*)+0x69f) [0x4a670f]
14: (MDS::_dispatch(Message*)+0x15c3) [0x4b6fa3]
15: (MDS::ms_dispatch(Message*)+0x57) [0x4b7617]
16: (SimpleMessenger::dispatch_entry()+0x7bb) [0x69f0fb]
17: (SimpleMessenger::DispatchThread::entry()+0x1c) [0x496e8c]
18: (()+0x7d60) [0x7f8f48f4cd60]
19: (clone()+0x6d) [0x7f8f47b7ae2d]
- Target version set to v0.32
- Assignee set to Yehuda Sadeh
- Assignee deleted (
Yehuda Sadeh)
- Status changed from New to In Progress
- Assignee set to Greg Farnum
- Target version changed from v0.32 to v0.33
I looked at this some Friday; I'm afraid that without the entire log or the on-disk state there's little chance of figuring out what went wrong here. Will discuss with Sage, but otherwise I think I need to close this until somebody manages to reproduce.
- Status changed from In Progress to Can't reproduce
Yep; the inode table was somehow not being read off disk properly, but we've no idea what's going on, exactly.
- Target version deleted (
v0.33)
- Project changed from Ceph to CephFS
- Category deleted (
1)
Bulk updating project=ceph category=mds bugs so that I can remove the MDS category from the Ceph project to avoid confusion.
Also available in: Atom
PDF