OpenLava: compute hosts can fail to start `openlava-lim` service

[vlj91]

When many compute hosts join the cluster ring at once, some nodes can fail to start openstack-lim service:

Sep 29 20:04:18.872111 5521 3 33 setMyClusterName: unable to find the cluster file containing local host flight-156
Sep 29 20:04:18.872177 5521 3 33 setMyClusterName: Above fatal error(s) found.

Restarting the service kicks the node back into life. Possibly some kind of race condition?

Process to replicate:

• Start a cluster using the 2016.3rc6 template (professional edition)
• Select openlava scheduler type
• Launch 24+ nodes
• Every few nodes will appear as unavailable

[mjtko]

For the record, this is still present in 2017.1 (@andrewprovis encountered it yesterday).

Note that this is currently a low priority to fix though.