Description of problem: Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
See https://retrace.fedoraproject.org/faf/reports/2170288/ includes a rudimentary BT
I see there's some RDMA code in the stack trace. Any chance this is RDMA-specific, e.g. some kind of ordering/serialization assumption fulfilled by the socket code but not by the RDMA code? That wouldn't necessarily mean it's a flaw in the RDMA code or should be fixed there, but might provide a useful hint. Or maybe it's pure coincidence.
Do we have a reproducer for this?
No, _I_ don't have a reproducer. This is an automated ABRT report coming from boxes running community gluster installed from the CentOS Storage SIG repos. I'm simply relaying the report that gets forwarded to me as the Gluster packager in Fedora. (CentOS ABRTs get sent to Fedora.) Although as you can see at the link posted in Comment 1 it has occurred 67 times, so it seems like it should be easy to reproduce.
Kaleb - do you still see this happening? I'm looking for a core file (specifically the complete backtrace) and the test which is run to get to the crash by which we can begin the investigation.
Release 3.12 has been EOLd and this bug was still found to be in the NEW state, hence moving the version to mainline, to triage the same and take appropriate actions.
Doesn't have sufficient data to debug this. Closing it.