Bug #551
closedcfuse crash on quick mds restart
0%
Description
Program terminated with signal 11, Segmentation fault.
#0 0x00000000004704ad in Client::kick_flushing_caps (this=0xe56000, mds=<value optimized out>) at client/Client.cc:2373
2373 assert(cap->session == session);
(gdb) bt
#0 0x00000000004704ad in Client::kick_flushing_caps (this=0xe56000, mds=<value optimized out>) at client/Client.cc:2373
#1 0x0000000000494ece in Client::handle_mds_map (this=0xe56000, m=<value optimized out>) at client/Client.cc:1466
#2 0x000000000049d06b in Client::ms_dispatch (this=0xe56000, m=0x1d4da00) at client/Client.cc:1377
#3 0x000000000044ca93 in Messenger::ms_deliver_dispatch (this=0xe51000) at msg/Messenger.h:97
#4 SimpleMessenger::dispatch_entry (this=0xe51000) at msg/SimpleMessenger.cc:342
#5 0x00000000004455ac in SimpleMessenger::DispatchThread::entry (this=0xe51488) at msg/SimpleMessenger.h:570
#6 0x0000000000457ffa in Thread::_entry_func (arg=0xe56000) at ./common/Thread.h:39
#7 0x00007f1159e9173a in start_thread (arg=<value optimized out>) at pthread_create.c:300
#8 0x00007f1158e5069d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
#9 0x0000000000000000 in ?? ()
The MDS crashed in a test (on purpose, using the export kill) and I restarted it quickly. When it came up, the client crashed with this backtrace.
I'm not sure at the moment if the appropriate behavior is to kill the assert, to assert that the MDS IDs match, or to not call kick_flushing_caps in the case of an MDS restart.