[openstack-dev] Trying again on wait_for_compute in devstack

Matt Riedemann mriedemos at gmail.com
Wed Aug 2 15:04:03 UTC 2017


On 8/2/2017 6:17 AM, Sean Dague wrote:
> and we're not going
> to use it in multinode scenarios.

Why would you not run this in multinode scenarios? That's the only time 
this is really a problem, because in the single node case we're 
discovering and mapping the compute node late enough that it's not been 
a problem.

The main failures are in the multinode jobs:

https://bugs.launchpad.net/grenade/+bug/1708039

https://goo.gl/xXhW8r

The devstack change is also failing on XenServer CI and who knows how 
many other 3rd party CIs that don't run against devstack changes, but 
will explode once it merges and they are running on Cinder or Neutron 
changes. I've dropped the tags in the subject line of this email so that 
this gets broader awareness as this isn't really just going to impact 
nova and ironic jobs if third party CIs aren't setup to handle this.

Why don't we wait until after RC1 on 8/10 before doing this? We already 
broke the gate and lost at least 6-10 hours last week on the day of 
feature freeze because of this.

-- 

Thanks,

Matt



More information about the OpenStack-dev mailing list