** Impact ** Since Essex, all instances will have an eth0 MAC address in the range of FA:16:3E, which is near the end of the MAC address space.
When openvpn is started, a TAP interface is created with a random generated MAC address. Chances are high the generated MAC address is lower in value than the eth0 MAC address. Once the tap interface is added to the bridge interface, the bridge interface will no longer have the eth0 MAC address, but take over the TAP MAC address. This is a feature of the linux kernel, whereby a bridge interface will take the MAC address with the lowest value amongst its interfaces. After the ARP entries expire, this will result in the cloudpipe instance being no longer reachable. This fix, randomly generates a MAC address starting with FA:17:3E, which is greater than FA, and will thus ensure the brige will keep the eth0 MAC address. ** Development Fix ** This fix has been addressed in https://review.openstack.org/6327 and already fixed in quantal. ** Stable Fix ** This fix has been addresses in the stable/essex tree in https://review.openstack.org/6327 ** Regression Potential ** Minimal, it is a hardyly used feature in Openstack on Ubuntu. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/975043 Title: [SRU] Cloudpipe VPN instance can loose connectivity after starting openvpn To manage notifications about this bug go to: https://bugs.launchpad.net/nova/+bug/975043/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
