Closed buzztroll closed 12 years ago
This problem is not related to current-reservations.txt. The instance information is in the Derby database.
What happens is that the method called by nimbus-admin to list all instances ends up calling impls.WorkspaceHomeImpl:find. This method tries to destroy expired instances and throws an exception for those. Because this exception is thrown, instances are not included in the result of nimbus-admin.
Should we show expired instances that couldn't be destroyed in the nimbus-admin output?
This tends to happen when a VMM node fails and then comes back up. The VMs Nimbus started are gone, but Nimbus doesn't seem to realize that.
Specifically on Alamo right now (Nimbus 2.9), VM instance 959 is shown in current-reservations.txt, there are repeating entries like below in services.log:
2012-08-16 08:42:08,898 INFO impls.InstanceResourceImpl [pool-3-thread-2,remove:767] [NIMBUS-EVENT][id-959]: destroy begins 2012-08-16 08:42:14,902 INFO impls.InstanceResourceImpl [pool-3-thread-2,remove:767] [NIMBUS-EVENT][id-959]: destroy begins
But that instance doesn't appear in "nimbus-admin -l".