Ilya, What hypervisor do you have? CleanupNetworkRules command is invoked by SecurityGroupManager when clenaup SG for the expunged/crashed User vm. SG is supported on Xen/KVM machines only. I suspect that in your case SG mapping was created for the vm although your hypervisor doesn't support it. Check "security_group_vm_map" to see if there are references to the vms that fail to expunge.
And the consistent failure in clenaup process will retain the vm in Destroyed state forever. As the workaround, I would suggest to delete the mappings for the VM from the DB table, and then vm expunge thread should clean up vm successfully. Deleting the vm from the DB w/o going through proper cleanup process might be dangerous as there can be existing DB entries referencing the object you've removed (or marked as removed). -Alena. On 11/12/12 2:47 PM, "Musayev, Ilya" <imusa...@webmd.net> wrote: >I gave up on the wait, went ahead and deleted destroyed VM (which should >have been expunged) from DB. > >I was then able to delete the network.. > >Thanks >ilya > >-----Original Message----- >From: Musayev, Ilya [mailto:imusa...@webmd.net] >Sent: Monday, November 12, 2012 5:31 PM >To: cloudstack-dev@incubator.apache.org >Subject: RE: can i force expunge > >Yeah, that's less than idea - you wont recall what was done to release >those IPs? > >I also notice this in my log when it tries to expunge but fails: > >2012-11-12 17:28:12,995 DEBUG [agent.transport.Request] >(DirectAgent-416:null) Seq 31-1309671426: Processing: { Ans: , MgmtId: >345051904793, via: 31, Ver: v1, Flags: 10, >[{"UnsupportedAnswer":{"result":false,"details":"Unsupported command >issued:com.cloud.agent.api.CleanupNetworkRulesCmd. Are you sure you got >the right type of server?","wait":0}}] } > > > >-----Original Message----- >From: Caleb Call [mailto:calebc...@me.com] >Sent: Monday, November 12, 2012 4:57 PM >To: cloudstack-dev@incubator.apache.org >Subject: Re: can i force expunge > >Does it show your instances sitting in an expunge state? I've had it >before not show the instances in that state (they get cleaned-up, but the >IPs don't), but it still thinks IPs are used. Infact right now this is >the case on my management network. It thinks 3 or 4 extra IPs are >actually used then what really is being used. I've always had to go in >to the database and clean this up. > > >On Nov 12, 2012, at 2:51 PM, "Musayev, Ilya" <imusa...@webmd.net> wrote: > >> Caleb >> >> The network gc is already set to 600seconds - which I think is pretty >>low. >> >> The ms has been restarted.. >> >> Thanks >> ilya >> >> -----Original Message----- >> From: Caleb Call [mailto:calebc...@me.com] >> Sent: Monday, November 12, 2012 4:27 PM >> To: cloudstack-dev@incubator.apache.org >> Subject: Re: can i force expunge >> >> Make sure you also drop the network.gc.interval and network.gc.wait in >>addition to expunge.delay and expunge.interval. Also, this may be >>obvious but you also need to restart the ms for the new intervals to >>take affect. >> >> Thanks >> >> On Nov 12, 2012, at 2:11 PM, "Musayev, Ilya" <imusa...@webmd.net> wrote: >> >>> Can I force expunge via API call or some other way? >>> >>> I lowered the expunge parameters in the GUI to 1800 seconds and 2 >>>threads, but it hasn't worked for me. >>> >>> I have vms that are destroyed but not expunged - that are holding the >>>IP I need to release to delete the network. >>> >>> Thanks >>> ilya >> >> >> > > > > > >
