Red Hat Bugzilla – Bug 809368
[ovirt] [vdsm] deadlock on SPM stop command
Last modified: 2016-02-10 11:28:31 EST
Description of problem:
I seem to hit a deadlock on vdsm after spmStop task fails on resource time-out.
I noticed that after this failure, host fails to handle connectStorageServer calls from backend, meaning, connectStorageServer is sent, iscsi login is performed,
but command seem to stuck on _invalidateAllPvs, and vdsm doesn't return valid response to backend.
Thread-277736:EBUG::2012-04-01 05:16:07,103::lvm::457::OperationMutex:_invalidateAllPvs) Operation 'lvm reload operation' is holding the operation mutex, waiting...
Thread-277738:EBUG::2012-04-01 05:16:07,826::BindingXMLRPC::167::vds:wrapper) [10.35.97.3
all problems started after failure in 'Thread-277609'.
in gdb, there are lots of threads waiting in cond.wait().
attached full gdb trace of all running threads, and vdsm log.
please note that it happened while connection connection to NFS server hosting 2 mount points was blocked.
Created attachment 574784 [details]
Closing old bugs. If this issue is still relevant/important in current version, please re-open the bug.