So, how masakari-hostmonitor works. It listens on {{ masakari_monitor_corosync_multicast_interface }} on each compute for traffic send by corosync regarding cluster status.
When it sees that corosync detects another host as down, it will send a message to engine.
So first of all I would check that pacemaker cluster is healthy and see host being down. You can use 'crm_mon -1' for that.
Next you should ensure that hostnames in corosync cluster is exactly the same as added to masakari API.
And of course check the masakari-hostmonitor logs on other computes
There was another issue in masakari itself, but it should be fixed in Xena, and message would be produced anyway in case of hitting it, so I think it's not that. But issue was when compute name was not matching between "openstack hypervisor list" and " openstack compute service list"
So, how masakari- hostmonitor works. It listens on {{ masakari_ monitor_ corosync_ multicast_ interface }} on each compute for traffic send by corosync regarding cluster status.
When it sees that corosync detects another host as down, it will send a message to engine.
So first of all I would check that pacemaker cluster is healthy and see host being down. You can use 'crm_mon -1' for that. hostmonitor logs on other computes
Next you should ensure that hostnames in corosync cluster is exactly the same as added to masakari API.
And of course check the masakari-
There was another issue in masakari itself, but it should be fixed in Xena, and message would be produced anyway in case of hitting it, so I think it's not that. But issue was when compute name was not matching between "openstack hypervisor list" and " openstack compute service list"