Linux bridge agent crashes without logging if l2pop is off and vxlan_group is unset

[Assaf Muller]

In the context of https://bugs.launchpad.net/openstack-ansible/+bug/1521793, patch https://review.openstack.org/#/c/253606/ was sent to test openstack-ansible deployment with LB agent and l2pop disabled. VMs aren't spawning, you can see binding failures in the neutron-server log. Finally, looking at the LB agent log it looks oddly short:

http://logs.openstack.org/06/253606/1/check/gate-openstack-ansible-dsvm-commit/187ccc0/logs/aio1_neutron_agents_container-8702df06/neutron-linuxbridge-agent.log

Output from that log:

2015-12-04 18:34:01.455 3062 DEBUG oslo_service.service [-] ******************************************************************************** log_opt_values /openstack/venvs/neutron-master/lib/python2.7/site-packages/oslo_config/cfg.py:2270
2015-12-04 18:34:01.455 3062 DEBUG neutron.agent.linux.utils [-] Running command: ['ip', '-o', 'link', 'show', 'eth12'] create_process /openstack/venvs/neutron-master/lib/python2.7/site-packages/neutron/agent/linux/utils.py:83
2015-12-04 18:34:01.486 3062 DEBUG neutron.agent.linux.utils [-]
Command: ['ip', '-o', 'link', 'show', 'eth12']
Exit code: 0
 execute /openstack/venvs/neutron-master/lib/python2.7/site-packages/neutron/agent/linux/utils.py:154
2015-12-04 18:34:01.488 3062 DEBUG neutron.agent.linux.utils [-] Running command: ['ip', '-o', 'link', 'show', 'eth11'] create_process /openstack/venvs/neutron-master/lib/python2.7/site-packages/neutron/agent/linux/utils.py:83
2015-12-04 18:34:01.523 3062 DEBUG neutron.agent.linux.utils [-]
Command: ['ip', '-o', 'link', 'show', 'eth11']
Exit code: 0
 execute /openstack/venvs/neutron-master/lib/python2.7/site-packages/neutron/agent/linux/utils.py:154
2015-12-04 18:34:01.524 3062 DEBUG neutron.agent.linux.utils [-] Running command: ['ip', 'addr', 'show', 'to', '172.29.242.170'] create_process /openstack/venvs/neutron-master/lib/python2.7/site-packages/neutron/agent/linux/utils.py:83
2015-12-04 18:34:01.557 3062 DEBUG neutron.agent.linux.utils [-]
Command: ['ip', 'addr', 'show', 'to', '172.29.242.170']
Exit code: 0
 execute /openstack/venvs/neutron-master/lib/python2.7/site-packages/neutron/agent/linux/utils.py:154
2015-12-04 18:34:01.558 3062 WARNING neutron.plugins.ml2.drivers.linuxbridge.agent.linuxbridge_neutron_agent [-] VXLAN muticast group(s) must be provided in vxlan_group option to enable VXLAN MCAST mode

And that's the end of the file. Looking at the code it's clear what's happening. Both l2pop and VXLAN.vxlan_group are off, so when the LB agent initializes and calls check_vxlan_support, it throws VxlanNetworkUnsupported, but that exception is uncaught and the agent crashes without logging anything.

[Assaf Muller (amuller)]

Assigning to Sean, our resident LB agent fan.

[Sean M. Collins (scollins)]

I'm marking this as a dup of https://bugs.launchpad.net/openstack-ansible/+bug/1521793 so we can tie everything together. There is probably work that needs to be done to make the LB agent catch the error, emit a useful diagnostic message and exit cleanly