Bug #70
Interrupting rados tool crashes osd?
0%
Description
432 16 2096 2080 19.2568 24 3.14879 3.31748
^C
gregf@pudgy:~/ceph/src$ 30658
30658 Process terminating with default action of signal 11 (SIGSEGV): dumpin
g core
30658 Access not within mapped region at address 0xAD
30658 at 0x45898E: T.2657 (Mutex.h:94)
30658 by 0x460084: SimpleMessenger::submit_message(Message*, SimpleMessen
ger::Pipe*) (SimpleMessenger.cc:2276)
30658 by 0x460475: SimpleMessenger::send_message(Message*, Connection*) (
SimpleMessenger.cc:411)
30658 by 0x4735C2: ReplicatedPG::eval_repop(ReplicatedPG::RepGather*) (Re
plicatedPG.cc:1908)
30658 by 0x4743CA: ReplicatedPG::op_commit(ReplicatedPG::RepGather*) (Rep
licatedPG.cc:1876)
30658 by 0x4A4E10: C_OSD_OpCommit::finish(int) (ReplicatedPG.cc:1760)
30658 by 0x5A67F7: Finisher::finisher_thread_entry() (Finisher.cc:54)
30658 by 0x4672A9: Thread::_entry_func(void*) (Thread.h:39)
30658 by 0x502D739: start_thread (pthread_create.c:300)
30658 by 0x5E8269C: clone (clone.S:112)
30658 If you believe this happened as a result of a stack
30658 overflow in your program's main thread (unlikely but
30658 possible), you can try to increase the size of the
30658 main thread stack using the --main-stacksize= flag.
30658 The main thread stack size used in this run was 8388608.
30658
This happened while running valgrind massif, but I think I saw it without any special tools too. Probably rados just isn't behaving well while shutting down or something.
Of course, a misbehaving client shouldn't be able to do that. Figure out what happened, and stop it!
History
#1 Updated by Sage Weil almost 14 years ago
- Status changed from New to Resolved
caused by sending message on closed Connection with null pipe.
#2 Updated by Sage Weil almost 14 years ago
- Status changed from Resolved to Closed