[monit] openvswitch and openvswitch-agent not being monitored on controller node

[jameswthorne]

On Ubuntu, monit is not monitoring the following on the controller node, but is on the compute node:

openvswitch-switch_ovsdb_server
openvswitch-switch_ovs_vswitchd
neutron-plugin-openvswitch-agent

On CentOS, monit is not monitoring the following on the controller node, but is on the compute node:

openvswitch_ovsdb_server
openvswitch_ovs_vswitchd
neutron-openvswitch-agent

This looks to have partly been fixed in #614 but only for Quantum.

[claco]

@jameswthorne Running in slow mode this morning. Those services aren't normally running on hacontroller1/2, at least in my fresh 4.2.2 kick w/ neutron as my provider. Can you elaborate on your setup, what you see on the controller?

[claco]

@jameswthorne Ah. I assume this is the same case. Since you're doing single-controller, single-network-node, we're dropping in the ovs plugn, but the monitoring check is out of date s/quantum/neutron. In hacontroller, I believe the monitoring is different.

So, seems like in the case of hac1/2, the monitoring is complete and correct right? We just now need to fix that recipe to drop things in single-controller/single-network-node properly.

[claco]

@jameswthorne I've fixed the issue in #614 which seems unrelated. Also confirmed that in the recommended hacontroller1 role configuration, monitoring for these services is dropped on the computes on both distros and not on the controller as the controllers don't perform that role in the ha setup. Since our installs are supposed to be hacontroller1 (even in instances where the aren't going to have a hacontroller2 right away), Going to leave the rest of this as a won't fix. Doesn't seem work the risk to shuffle the roles around at this point.