nova compute manager traceback during _sync_power_states

[Andrew Glen-Young]

After upgrading nova-compute this morning to the latest nova-compute package in Precise I have noticed a non-fatal traceback in /var/log/nova/nova-compute.log.

Traceback:

2012-02-22 15:04:28,413 ERROR nova.manager [-] Error during ComputeManager._sync_power_states: 'dict' object has no attribute 'state'
(nova.manager): TRACE: Traceback (most recent call last):
(nova.manager): TRACE: File "/usr/lib/python2.7/dist-packages/nova/manager.py", line 155, in periodic_tasks
(nova.manager): TRACE: task(self, context)
(nova.manager): TRACE: File "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 2143, in _sync_power_states
(nova.manager): TRACE: vm_power_state = vm_instance.state
(nova.manager): TRACE: AttributeError: 'dict' object has no attribute 'state'
(nova.manager): TRACE:

This does not seem to affect the nova-compute daemon (so far).

$ dpkg-query --show nova-compute
nova-compute 2012.1~e4~20120217.12709-0ubuntu1

$ cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=12.04
DISTRIB_CODENAME=precise
DISTRIB_DESCRIPTION="Ubuntu precise (development branch)"

[Adam Gandelman (gandelman-a)]

Hi Andrew, thanks for the report. To be clear, was this an upgrade from the Oneiric/Diablo to Preicse/Essex, or an upgrade between version of Essex? Also, is it correct to assume that, at the time of the upgrade, there were instances running on the compute node? Thanks

[Andrew Glen-Young (aglenyoung)]

@Adam:

The upgrade was an upgrade of the Ubuntu packages from the Precise/Essex (i.e. _not_ between Openstack releases). There were instances running at the time.

[Thierry Carrez (ttx)]

@Andrew: there were quite a few changes around get_info over the last 10 days... and the way it looks currently it should just work... any chance you could reproduce on a more recent build ?

[Andrew Glen-Young (aglenyoung)]

@Thierry: I believe that new(er) Precise packages should be uploaded today (Friday). I'll test them early next week and let you know if I see this error.

[Dan Prince (dan-prince)]

This is a dup of 934427.

https://bugs.launchpad.net/nova/+bug/934276

The should be pretty simple:

+++ b/nova/compute/manager.py
@@ -2244,7 +2244,7 @@ class ComputeManager(manager.SchedulerDependentManager):
             db_power_state = db_instance['power_state']
             try:
                 vm_instance = self.driver.get_info(db_instance)
- vm_power_state = vm_instance.state
+ vm_power_state = vm_instance['state']