[ovs-discuss] Management interface not accessible after reboot

[Gurucharan Shetty]

On Mon, Dec 8, 2014 at 1:58 AM, Emil Flink <emil.flink at gmail.com> wrote:
> Hi,
>
> I'm doing some testing with Open vSwitch 2.3.0 on Debian 7 (stable) compiled
> from source and have encountered a fairly weird error. I don't know if it is
> a problem with Debian or Open vSwitch.
>
> The mgmt0 interface comes up fully configured after reboot but is not
> accessible until I have "used" the network from the server, ie. by pinging
> another host.
When you say it is not accessible, I suppose you mean that you cannot
ping it from outside? If so, I would put tcpdump on your physical
interfaces to see whether any request is even coming to your ethernet
interfaces. I would suggest removing the bond and attach to a single
physical interface to simplify your network configuration and see
whether the problem still exists.

>
> The very first ping will have a high ms of 1000+, indicating something is
> going on in the background at this point.
>
>
> My /etc/network/interfaces configuration:
>
> auto lo0
> iface lo inet loopback
>
> allow-vmbr0 vbond0
> iface vbond0 inet manual
>     ovs_bridge vmbr0
>     ovs_type OVSBond
>     ovs_bonds eth0 eth1
>     pre-up ( ifconfig eth0 mtu 9000 && ifconfig eth1 mtu 9000 )
>     post-up ( ifconfig vbond0 mtu 9000 )
>     ovs_options bond_mode=active-backup
>
> allow-ovs vmbr0
> iface vmbr0 inet manual
>     ovs_type OVSBridge
>     ovs_ports vbond0 mgmt0
>     mtu 9000
>
> allow-vmbr0 mgmt0
> iface mgmt0 inet static
>     ovs_bridge vmbr0
>     ovs_type OVSIntPort
>     address 10.69.52.6
>     netmask 255.255.255.0
>     gateway 10.69.52.1
>     dns-nameservers 10.69.8.254
>     mtu 1500
>
> _______________________________________________
> discuss mailing list
> discuss at openvswitch.org
> http://openvswitch.org/mailman/listinfo/discuss
>