]> git.apps.os.sepia.ceph.com Git - ceph.git/commit
mds,qa: exit instead of respawn under valgrind
authorPatrick Donnelly <pdonnell@redhat.com>
Wed, 3 Mar 2021 03:39:09 +0000 (19:39 -0800)
committerPatrick Donnelly <pdonnell@redhat.com>
Wed, 3 Mar 2021 17:30:21 +0000 (09:30 -0800)
commit5faf0ee0f367f66e25a3927e6ac4c40fb83568ce
treeaa81e01575e60a3932494339997cb05f3a8eee2a
parent3e5e03d4d29020a918fd34591e9cdd78c38579e4
mds,qa: exit instead of respawn under valgrind

valgrind can't handle execve of /proc/self/exe:

    2021-02-27T05:52:37.813 INFO:tasks.ceph.mds.d.smithi073.stderr:==00:01:03:20.556 41218== execve(0x18546740(/proc/self/exe), 0x18546670, 0x133ef310) failed, errno 2
    2021-02-27T05:52:37.813 INFO:tasks.ceph.mds.d.smithi073.stderr:==00:01:03:20.556 41218== EXEC FAILED: I can't recover from execve() failing, so I'm dying.
    2021-02-27T05:52:37.813 INFO:tasks.ceph.mds.d.smithi073.stderr:==00:01:03:20.556 41218== Add more stringent tests in PRE(sys_execve), or work out how to recover.

So configure the MDS to just exit so it can be restarted by QA infra (the
daemon watchdog).

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
qa/suites/fs/verify/validater/valgrind.yaml
qa/tasks/daemonwatchdog.py
src/common/options.cc
src/mds/MDSDaemon.cc