Red Hat Bugzilla – Bug 961133
OVS L2 Agent looks for bridge changes via inefficient polling
Last modified: 2016-04-26 18:35:37 EDT
See upstream bug: https://bugs.launchpad.net/quantum/+bug/1177973
We need info how to test it on non dev stack setup , is looking on CPU is enough ?
Ofer: Validation could include detecting the invocation rate for ovs-vsctl (the command which polling minimization affects) , since the cpu load would vary by the performance of a given system. I would also suggest killing the monitor process and ensuring that the agent falls back to polling (if no respawn interval is provided) or respawns the monitor process (if respawn interval is provided).
The upstream didn't merge in time for Havana, and therefore doesn't have to be considered for 4.0.