On Fri, Jul 14, 2017 at 12:34:44PM +0530, Numan Siddique wrote:
> Hi Han,
>
> I think I am seeing a similar issue here [1]. Can you please check and see
> if it's the same issue hit here which this patch fixes.
>
> [1] uses OVS 2.7.1 for tripleo and the upgrage job updates OVS 2.6.1 to
> OVS 2.7.1. The restart of openvswitch service fails and there is a crash in
> ovs-vswitchd. Please see the end of the log file (it's not too big).
>
> [1] -
> http://logs.openstack.org/91/473191/12/check/gate-tripleo-ci-centos-7-multinode-upgrades-nv/414af3c/logs/subnode-2/openvswitch/ovs-vswitchd.txt.gz
> https://review.openstack.org/#/c/473191/
>
> 2017-07-14T04:14:04.229Z|00038|in_band|ERR|br-ex: failed to initialize
> in-band control: cannot open datapath local port br-ex (File exists)
> 2017-07-14T04:14:04.285Z|00002|daemon_unix|ERR|fork child died before
> signaling startup (killed (Segmentation fault), core dumped)
> 2017-07-14T04:14:04.285Z|00003|daemon_unix|EMER|could not detach from
> foreground session
>
> This segfault is consistent. I did 2 rechecks and I see the same issue.
The ultimate cause of the segfault is the new netdev_open() behavior,
but the proximate cause of this segfault is a separate bug that we can
fix easily:
https://mail.openvswitch.org/pipermail/ovs-dev/2017-July/335782.html
Once the proximate cause is fixed, we probably need to reassess what
you're reporting.
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
