Description of problem:


If you set this:

CONFIG_NEUTRON_OVS_BRIDGE_IFACES=br-ex:enp10s0

your external interface will be connected into br-ex (provider network bridge), and neutron now sets the bridge in secure mode.

That means, that after reboot, the flows of such flow will be completely empty, leading to the node being able to communicate with the outside world, compute nodes, etc.


This can be fixed by changing /etc/sysconfig/network-scripts/ifcfg-br-ex


   OVS_EXTRA="set bridge br-ex other-config:hwaddr=8c:89:a5:1a:a8:65"

   to 

   OVS_EXTRA="set bridge br-ex other-config:hwaddr=8c:89:a5:1a:a8:65 fail_mode=standalone"


How reproducible:

Always.

Steps to Reproduce:
1. Deploy with packstack, mapping the main interface to a provider network bridge
2. [Update to RHEL 7.3] - we are unsure about this step being necessary.
3. Reboot

Actual results:

The node takes a long time to boot, and does not get connectivity. This is more clear to happen if your network connectivity works with DHCP. (OVSBOOTPROTO="dhcp")


Expected results:

Normal boot.


Additional info:

This is related to this bug assigned to osp-director: https://bugzilla.redhat.com/show_bug.cgi?id=1386299