Description of problem:
tobiko.tests.faults.ha.test_cloud_recovery.DisruptTripleoNodesTest.test_hard_reboot_controllers_recovery has been failing occasionally [1]

It appears that neutron_l3_agent is not in the list of running containers when a controller gets rebooted. I inspected tobiko log file for the test_hard_reboot_controllers_recovery test section [2].
However, the controller-2 l3-agent log does not show any abnormality [3] in fact it shows it as running and spinning off haproxy side containers which can be seen in the tobiko log list.
This is as far as I took this debug, we need to see if this is perhaps a test problem, i.e. not getting an accurate list of running containers. 
OR this could be related to an existing BZ2133743. In that BZ, the haproxy container is not started but not the l3-agent container is.

[1] https://rhos-ci-jenkins.lab.eng.tlv2.redhat.com/view/DFG/view/network/view/neutron/job/DFG-network-neutron-16.2_director-rhel-virthost-3cont_2comp-ml2ovs-ipv4-vxlan-tobiko-tempest-dvr/35/testReport/junit/tobiko.tests.faults.ha.test_cloud_recovery/DisruptTripleoNodesTest/Tobiko___test_hard_reboot_controllers_recovery/

[2] timestamp 23:34:15(first output) http://rhos-ci-logs.lab.eng.tlv2.redhat.com/logs/rcj/DFG-network-neutron-16.2_director-rhel-virthost-3cont_2comp-ml2ovs-ipv4-vxlan-tobiko-tempest-dvr/35/test_results/tobiko_faults-neutron_1/tobiko.log

[3] http://rhos-ci-logs.lab.eng.tlv2.redhat.com/logs/rcj/DFG-network-neutron-16.2_director-rhel-virthost-3cont_2comp-ml2ovs-ipv4-vxlan-tobiko-tempest-dvr/35/controller-2/var/log/containers/neutron/l3-agent.log.1.gz