Ask Your Question
1

"nova agent-list" returns empty

asked 2014-05-23 17:56:13 -0500

NULL gravatar image

Hi, I was trying to configuration OpenStack Icehouse Ubuntu distro following instructions in the official installation manual. It contains three nodes: controller, compute1 and network.

When I tried to create an instance, it ran into ERROR state. Relevant logs are attached below:

$ nova console-log 1c65d073-b9fe-4df6-922e-dee8e1f5fc15
ERROR: Instance 1c65d073-b9fe-4df6-922e-dee8e1f5fc15 is not ready (HTTP 409) (Request-ID: req-23631d2b-9578-4b67-bb6c-822d1286c4cb)

nova/nova-scheduler.log:
2014-05-23 15:15:06.125 1042 WARNING nova.scheduler.driver [req-8f37769b-60e9-48f1-849c-e47055948942 e4602cc5b7f941b8b526cefca2e600b7 53a432f92a174b7bbf7ac21c1e93bc8d] [instance: 1c65d073-b9fe-4df6-922e-dee8e1f5fc15] Setting instance to ERROR state.

What surprises me is that "nova agent-list" command returns an empty list of computation agents. Does it mean that my compute1 node is not connected to controller at all? Because I grep the log folder in compute1 node and there is no log containing the instance ID. So I guess it hasn't received the instance-creation instruction from the controller.

But then how can I debug this situation? Is there any further detailed log about why nova-scheduler put the instance into ERROR state?

Thanks!

edit retag flag offensive close merge delete

Comments

During instance start up does it complain about address 169.254.169.254 ?
What is status of neutron-* services ?

dbaxps gravatar imagedbaxps ( 2014-05-24 00:46:32 -0500 )edit

@dbaxps No, it doesn't complain address or anything...and neutron services is good and running. neutron agent-list seems good for the network node and compute1 node.

I was just wondering if any detailed logs could be found for nova-scheduler service or how can I enable them.

NULL gravatar imageNULL ( 2014-05-24 11:59:13 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
1

answered 2014-05-24 16:28:33 -0500

craigtracey gravatar image

There are many reasons why the Nova scheduler would put an instance into an error state. Most common, in my opinion, would be that there is no valid host to schedule it to, but your mileage may vary. The best way to determine why an instance is in the error state is to run "nova show <uuid>". This should give you the error that was encountered. Similarly, the nova-scheduler log will clearly state why it could not proceed as expected.

As for "nova agent-list" being empty, what you really want to be running is "nova service-list." This will give you the current status of the various nova services within the cluster.

Hope this helps, Craig

edit flag offensive delete link more

Comments

Thanks Craig. Actually I found the cause to be that each time server reboots, rabbitmq reset its password back to the default value 'guest'. Now I have specified the password in rabbitmq.conf file. Things all right now! :)

And you are right, "nova agent-list" doesn't always mean computation node.

NULL gravatar imageNULL ( 2014-05-27 17:45:07 -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

3 followers

Stats

Asked: 2014-05-23 17:56:13 -0500

Seen: 2,981 times

Last updated: May 24 '14