[Crowbar] Stuck on 'build' status
csanburn at redwoodit.com
Thu Jan 12 10:22:32 CST 2012
Here's the general procedure I followed while trying to launch instances:
Logged into each nova-compute node and cleaned out all the nova logs, just to make it easier on myself. Less information to dig through
(Note, I didn't do it, but might be helpful to clear out the logs in the nova multi controller node as well, so you can monitor those)
I then went to the Nova Dashboard, already changed my default security group to allow icmp, ssh and 3389 (3389 for the Windows instances I was testing)
In Nova Dashboard I'd then switch to System Panel -> Instances and see what host it had picked to launch the instance on.
Log into that host and monitor the nova logs.
I had 3 nova-compute nodes so if the first one got stuck at the build status I'd just leave it there and launch another instance. I'd go back to the system panel to see which of the remaining nodes received that instance and repeat my monitoring. I'd do that until I had an instance on each node. In one instance I found that only one node could bring an instance to active, the other 2 got stuck on build. I rebooted both, got all services back up and running and then found I could get them active on all 3 nodes.
I never figured out what was exactly wrong. I'd find messages in logs that suggested I make sure that virtual support was active in my CPU. I checked this when I rebooted each node and found that support was already on for both of them.
If nothing else, at least this procedure will test all of the compute nodes to see if it's isolated to just one or two.
From: crowbar-bounces at dell.com [mailto:crowbar-bounces at dell.com] On Behalf Of Kevin Bringard
Sent: Thursday, January 12, 2012 11:06 AM
Cc: crowbar at lists.us.dell.com
Subject: Re: [Crowbar] Stuck on 'build' status
I'd start by checking the nova-scheduler log (on the controller). That'll show you an error, or will tell you which compute it is trying to cast it to. Then you can go there and check the nova-compute log for errors. Should be a good place to start.
On Jan 12, 2012, at 9:03 AM, "Paul Regan" <me at paulregan.co.uk> wrote:
> Sorry bad description. I think the default image is supplied by
> Crowbar - ubuntu-11.04-server. Its this I'm creating an instance from
> The test environment is on metal rather than VMware ..
> On 12 January 2012 15:43, <Rob_Hirschfeld at dell.com> wrote:
>> Default image in Nova? If you are using VMs for Crowbar, did you change to Nova to use Qemu?
>> -----Original Message-----
>> From: crowbar-bounces On Behalf Of Paul Regan
>> Sent: Thursday, January 12, 2012 9:39 AM
>> To: crowbar
>> Subject: [Crowbar] Stuck on 'build' status
>> I appear to have a working Crowbar install, all green and services in the console are all OK (excluding nova-vncproxy, which I understand is known).
>> Creating an instance based on the default supplied Ubuntu image but it sits in 'Build' state. I have been following the other thread thats been running on this topic which has given me pointers but no fix.
>> I've crawled all over the logs on all the servers (its a small POC
>> lab) and nothings jumping out. Lots of RabbitMQ unreachable errors, although it appears up (with my limited MQ knowledge) and active in nagios so maybe unrelated.
>>>> 2012-01-12 09:31:42,927 ERROR nova.rpc [-] AMQP server on 192.168.124.81:5672 is unreachable: [Errno 111] ECONNREFUSED. Trying again in 3 seconds.
>>>> 2012-01-12 09:31:45,971 INFO nova.rpc [-] Connected to AMQP server
>>>> 2012-01-12 09:31:45,996 DEBUG nova.rpc [-] Re-established AMQP
>>>> queues from (pid=1890) reconnect
>> I guess I'm looking for pointers on where to look, or what to try next ..
>> Crowbar mailing list
>> Crowbar at dell.com
>> For more information: https://github.com/dellcloudedge/crowbar/wiki
> Crowbar mailing list
> Crowbar at dell.com
> For more information: https://github.com/dellcloudedge/crowbar/wiki
Crowbar mailing list
Crowbar at dell.com
For more information: https://github.com/dellcloudedge/crowbar/wiki
More information about the Crowbar