[ovs-discuss] ovs-vswitchd crash during startup when br-ex has assigned IP address and br-ex is UP

Ben Pfaff blp at ovn.org
Wed Aug 9 17:21:51 UTC 2017

On Wed, Aug 09, 2017 at 03:54:50PM +0000, Prokes, JiriX X wrote:
> We are using kolla/kolla-ansible to deploy openstack.
> Used configuration is ovs + vxlan + old (Opendaylight)
> Opendaylight needs to have IP address assigned to br-ex bridge to enable connectivity via floatingIP to VMs.
> Ovs is running in docker container.
> After deployment ovs container is up and running and br-ex is in state DOWN
> ip addr show br-ex
> 12: br-ex: <BROADCAST,MULTICAST> mtu 1500 qdisc noqueue state DOWN qlen 1000
>     link/ether 68:05:ca:32:62:93 brd ff:ff:ff:ff:ff:ff
> Then we assign IP and set it UP
> ip addr show br-ex
> 12: br-ex: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN qlen 1000
>     link/ether 68:05:ca:3a:19:33 brd ff:ff:ff:ff:ff:ff
>     inet scope global br-ex
>        valid_lft forever preferred_lft forever
>     inet6 fe80::6a05:caff:fe3a:1933/64 scope link
>        valid_lft forever preferred_lft forever
> Then ovs container is restarted
> docker restart openvswitch-vswitchd
> Which cause ovs-vswitchd crash.

Thank you for the report.

I believe that this is the bug fixed by the following commit.  This bug
fix is on master, in 2.7.2, and in the branch for the upcoming 2.8.0

commit 66a9ef7a87f77a25ed568f55c1789eb6075b6abe
Author: Ben Pfaff <blp at ovn.org>
Date:   Mon Jul 17 09:54:54 2017 -0700

    connmgr: Fix crash when in_band_create() fails.
    update_in_band_remotes() created an in-band manager and then tried to work
    with it without first checking whether creation had succeeded.  If it
    failed, this led to a segfault.
    Reported-by: Numan Siddique <nusiddiq at redhat.com>
    Reported-at: https://mail.openvswitch.org/pipermail/ovs-dev/2017-July/335530.html
    Signed-off-by: Ben Pfaff <blp at ovn.org>
    Acked-by: Justin Pettit <jpettit at ovn.org>

More information about the discuss mailing list