On Tue, 12 May 2020 at 10:47, Brad Hubbard <bhubbard@redhat.com> wrote:
This job only takes about 15 minutes to get to the point where it errors but then runs for several hours after that. I'd suggest running it again and inspecting the status of the daemons once you know the error has occurred.
I logged into the machine and checked for existence of /sys/fs/fuse/connection and the mountpoint and checked whether Ceph FS was mounted and finally, checked Ceph cluster's status. The results[1][2][3] were positive for all the checks, So I figured that the execution probably needs to wait a bit before running "ls /sys/fs/fuse/connections"[4] and it worked. The execution moved past that point and testsuite crashed at a different point. I am trying to find out exact cause, I'll mail on this thread in case I can't. Thanks for the help, Brad! [1] https://gist.github.com/rishabh-d-dave/eef6cdb21f54a95edec25d412e52d09e#file... [2] https://gist.github.com/rishabh-d-dave/eef6cdb21f54a95edec25d412e52d09e#file... [3] https://gist.github.com/rishabh-d-dave/eef6cdb21f54a95edec25d412e52d09e#file... [4] https://github.com/rishabh-d-dave/ceph/commit/325c7f0447112a90dea656c5296852...; see the commit with title "DNM: let's wait before checking connection dir" in case commit SHA changes