Description of problem:
When using IPv6 tenant networks, the compute node will detect the IPv6 router announcements (RAs) from Neutron, which poisons the IPv6 route tables.
Version-Release number of selected component (if applicable):
OSP 7.3 Beta
Steps to Reproduce:
1. Deploy overcloud
2. Create IPv6 tenant networks
3. Add a router to the network
4. Start an instance on the network
The qbr interface accepts the RA from the Neutron router, and the default route for IPv6 gets set to the Neutron router.
The compute nodes should ignore RAs on the OVS bridges
In order to turn this behavior on/off, you can edit /proc/sys/net/ipv6/conf/default/accept_ra to be "0" instead of "1".
You can also turn it on/off on a per-interface basis, e.g.:
echo "1" > /proc/sys/net/ipv6/conf/eth0/accept_ra
This needs to be tested, so we can determine if RAs will still be accepted on interfaces configured with IPv6, or if we need to set accept_ra to "1" for the individual interfaces.
There is a downstream review of the changes needed to fix this bug here:
I tested it, and it appears to work.
followup after irc chat just now about where this fix lands:
the fix for this bug is included in the fix for bug 1303758 (links to gerrit above the description) - from that bug I can see it is in "Fixed In Version: openstack-tripleo-heat-templates-0.8.6-117.el7ost"
[root@overcloud-compute-0 ~]# cat /proc/sys/net/ipv6/conf/default/accept_ra
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.