Ask Your Question

Problem debugging compute

asked 2013-12-19 00:01:53 -0500

Qrilka gravatar image

I have a test cluster (nodes running in VirtualBox) and after some playing with instances I received Nova Compute in invalid state:

root@node-1:~# nova-manage service list
Binary           Host                       Zone             Status     State Updated_At
nova-consoleauth node-1                     internal         enabled    :-)   2013-12-19 05:29:42
nova-cert        node-1                     internal         enabled    :-)   2013-12-19 05:29:41
nova-scheduler   node-1                     internal         enabled    :-)   2013-12-19 05:29:37
nova-conductor   node-1                     internal         enabled    :-)   2013-12-19 05:29:36
nova-network     node-2                     internal         enabled    :-)   2013-12-19 05:29:39
nova-compute     node-2                     nova             enabled    XXX   2013-12-18 08:45:51
nova-network     node-3                     internal         enabled    :-)   2013-12-19 05:29:43
nova-compute     node-3                     nova             enabled    XXX   2013-12-18 08:45:51

And in horizon I have currently 2 insances: 1 in "Scheduling" and other in "Deleting" state. I have set "debug=true" in nova.conf but not see anything in logs, e.g. after restartiing nova-compute on compute nodes I don't see any log messages in /varl/log/nova/nova-compute.log

Seems like I'm stuck and googling does not give me any new hints.

Are there any ways to find out what has happened to Compute?

edit retag flag offensive close merge delete


Look at the compute logs in all the compute nodes. Here it is scheduling state, look at the scheduler and conductor logs as well. Refer the following link for detail troubleshooting on the issue.

dheeru gravatar imagedheeru ( 2013-12-19 00:12:47 -0500 )edit

Your link gives 404, I could not find scheduler log and as I say above nova-compute.log on compute nodes have no meaningful info

Qrilka gravatar imageQrilka ( 2013-12-19 00:40:58 -0500 )edit

I suppose the link should point to ?

Qrilka gravatar imageQrilka ( 2013-12-19 00:46:53 -0500 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2013-12-19 00:42:23 -0500

Anand TS gravatar image

Hi ,

In this case you should bring your nova-compute in both nodes by running debug command.

#/usr/bin/nova-compute --config-file /etc/nova/nova.conf

Then reset the state of the instance to active mode as in below link, then try deleting the instance again.


Thanks, Anand

edit flag offensive delete link more


and what is the difference between manual start of `nov-compute` above and starting it though service? Command like looks the same, I'll look into instances resetting, thanks

Qrilka gravatar imageQrilka ( 2013-12-19 00:46:13 -0500 )edit

resetting helped to delete node in "Scheduling" state but node in "Deleting" stuck in "Deleting" once again

Qrilka gravatar imageQrilka ( 2013-12-19 00:53:24 -0500 )edit

Just check your openstack services again. Maybe compute service has gone down again. This is a temporary solution of making nova-compute service up by running debug command. If you close the debug cli, it goes to down state again.

Anand TS gravatar imageAnand TS ( 2013-12-19 01:02:28 -0500 )edit

Running `nova-compute` in CLI does not change the situation: I see no output to stdout. Both instance termination in horizion and `nova delete` in command line just hang. And after that `nova list` and horizon show instance in Active status and horizion also show "Deleting" task for it.

Qrilka gravatar imageQrilka ( 2013-12-19 01:23:31 -0500 )edit

I may help you if you give more details through email.

Anand TS gravatar imageAnand TS ( 2013-12-19 02:32:03 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2013-12-19 00:01:53 -0500

Seen: 443 times

Last updated: Dec 19 '13