Update:
Modifying those inactivity probe timers did not appear to make a difference. In a few different Stein environments we are seeing similar behavior. If the connection to OVS is lost, like so:
2020-04-02 14:50:21.372 3526056 WARNING ovsdbapp.backend.ovs_idl.vlog [-] tcp:127.0.0.1:6640: send error: Broken pipe 2020-04-02 14:50:41.987 3526056 ERROR OfctlService [-] unknown dpid 112915483458894 2020-04-02 14:50:42.050 3526056 WARNING ovsdbapp.backend.ovs_idl.vlog [-] tcp:127.0.0.1:6640: connection dropped (Broken pipe)
Then the agent will implement a drop flow and few other flows (with new cookies) similar to this:
https://bugs.launchpad.net/neutron/+bug/1864822/comments/11
Here are some agent log messages around the time of the issue:
http://paste.openstack.org/show/791823/
Update:
Modifying those inactivity probe timers did not appear to make a difference. In a few different Stein environments we are seeing similar behavior. If the connection to OVS is lost, like so:
2020-04-02 14:50:21.372 3526056 WARNING ovsdbapp. backend. ovs_idl. vlog [-] tcp:127.0.0.1:6640: send error: Broken pipe backend. ovs_idl. vlog [-] tcp:127.0.0.1:6640: connection dropped (Broken pipe)
2020-04-02 14:50:41.987 3526056 ERROR OfctlService [-] unknown dpid 112915483458894
2020-04-02 14:50:42.050 3526056 WARNING ovsdbapp.
Then the agent will implement a drop flow and few other flows (with new cookies) similar to this:
https:/ /bugs.launchpad .net/neutron/ +bug/1864822/ comments/ 11
Here are some agent log messages around the time of the issue:
http:// paste.openstack .org/show/ 791823/