[lxc-users] LXC with Openvswitch and MAC forwarding table

[Chris Bennett]

Hi there,

I'm using ovsup/ovsdown scripts for attaching an LXC container to
Openvswitch.

I'm finding that a reboot of the container will result in no flow of
traffic from external node to the container until one of the following
occurs:

  - the LXC container generates an egress packet out the eth/veth
    interface
  - ARP entry expires from physical router

What I'm observing is OVS's forwarding table shows the container's MAC
address with the 'wrong' port id.

  # ovs-appctl fdb/show br-inet | grep fe:dd:3f:1e:71:f6
      1     0  fe:dd:3f:1e:71:f6  140

After generating a single packet from inside the container causes OVS
to learn the MAC on the correct port.

  # ping -c 1 192.168.0.1
  PING 192.168.0.1 (192.168.0.1) 56(84) bytes of data.
  64 bytes from 192.168.0.1: icmp_seq=1 ttl=253 time=5.52 ms

  --- 192.168.0.1 ping statistics ---
  1 packets transmitted, 1 received, 0% packet loss, time 0ms
  rtt min/avg/max/mdev = 5.528/5.528/5.528/0.000 ms



  # ovs-appctl fdb/show br-inet | grep fe:dd:3f:1e:71:f6
     12     0  fe:dd:3f:1e:71:f6    0

This might be more of a OVS mailing list question but I guess I'm
looking for a LXC user that has seen similar symptoms and might know
of how I can resolve this.  Maybe there is a way of seeding the OVS
forwarding table via the lxc.network.script.up script since the MAC is
known in advance by the LXC config?

Thanks,

Chris