I did a fresh install using lucid and the same thing happens with one exception. On oneiric, the AMQP trace showing in the log is always associated with a vm never transitioning from pending to running. On lucid, I still sometimes see the error in the log but the VM does come up.
There must be some kind of race condition. Is there currently a test for nova that
1. Deploys a multinode configuration
2. Starts and terminates groups of vms for a long time, making sure they all end up running
3. Also fails if any error is written to a log file
Also, I did the latest install using ppa:openstack-release/2011.3 but I sometimes see things like this in the log. I find these messages in bug tickets that seem like they were fixed for diablo so I am not sure why I see them:
DBError: exceptions must be old-style classes or derived from BaseException, not NoneType
I am using mysql. In case it matters I am running nova-network on both nodes and also nova-api on both nodes as suggested by you here http:// osdir.com/ ml/openstack- cloud-computing /2011-09/ msg00283. html
I did a fresh install using lucid and the same thing happens with one exception. On oneiric, the AMQP trace showing in the log is always associated with a vm never transitioning from pending to running. On lucid, I still sometimes see the error in the log but the VM does come up.
There must be some kind of race condition. Is there currently a test for nova that
1. Deploys a multinode configuration
2. Starts and terminates groups of vms for a long time, making sure they all end up running
3. Also fails if any error is written to a log file
Also, I did the latest install using ppa:openstack- release/ 2011.3 but I sometimes see things like this in the log. I find these messages in bug tickets that seem like they were fixed for diablo so I am not sure why I see them:
DBError: exceptions must be old-style classes or derived from BaseException, not NoneType