Open Stack

On 07/11/2013 01:37 PM, Nachi Ueno wrote:
> Hi Sean
>
> Sorry for it taking long time to fixing this problem.
> At least, 3 neutron core dev is working on this issue, but
> it is kind of timing issue so we are struggling to replicate it.
>
> I'm also OK to move it for non-voting now.

Great. I think the focus really should be to ensure that Neutron is 
generating enough and the right logs so that we can do first fail 
detection in the gate.

This really should be debuggable completely based on gate logs, and if 
not, we should increase what Neutron is printing out. For most Nova 
flakey fails we can get very close to the root cause based on the logs 
collected, which means recreating locally isn't really the path to 
success, but using the gate itself.

If anyone needs help sifting on the gate, please just ask for it on 
#openstack-dev, -infra, or -qa and we'll get people sorted.

	-Sean

-- 
Sean Dague
http://dague.net