Description of problem: neutron-openvswitch-agent stops after receiving SIGTERM. In the event of a sigsev on openvswitch, systemd based on to the dependencies defined send the SIGTERM signal to neutron-openvswitch-agent. The dataplane and controlplane get disrupted. Version-Release number of selected component (if applicable): RDO newton openstack-neutron.noarch 1:9.3.2-0.20170421152919.6b11276.el7.centos python-openvswitch.noarch 1:2.6.1-4.1.git20161206.el7 openvswitch.x86_64 1:2.6.1-4.1.git20161206.el7 openstack-neutron-ml2.noarch 1:9.3.2-0.20170421152919.6b11276.el7.centos How reproducible: Always Steps to Reproduce: 1. Kill ovs-vswitchd process 2. 3. Actual results: Both openvswitch and neutron-openvswitch-agent are in inactive status Expected results: systemd to handle the failure and restart the services Additional info: This is already fixed in openvswitch 2.7 [1] [1] https://github.com/openvswitch/ovs/commit/c19bf36d848cbdf755c6760fad1726c95e4377f1
This is fixed in openvswitch-2.6.1-11.git20161206.el7ost. We need to ensure RDO Newton and Ocata (Both on OVS 2.6) should be using that version.
Newton is EOL long ago so closing this bug as EOL, ocata+ have openvswitch-2.9.0+.