[Rdo-list] [tripleo] [oooq] Scaling out fails for overcloud deploy

[Gerard Braad]

Hi,


Currently trying to scale out using a TripleO-quickstart setup. I have
created an environment with 3 control nodes and 9 computer nodes using
a config.yaml.

As extra_args in this configuration I purposely only did
--control-scale 3 (with pacemaker). The deploy of --tags all finishes
completely. And as expected, only 1 (out of 9) compute node is up.

As a result I have an environment that is working as
  * undercloud
  * control-[0-2]
  * compute-7

After this, I login to the undercloud machine. Check the status with
ironic node-list and see all nodes in Provisiong State = available,
all remaining compute nodes show Power State = None and the other
nodes are seen as power off (control-[0-2] and compute-7).

Now when retrying to deploy with e.g. `--compute-scale 2` the
overcloud can not find request nodes to deploy to. The stack fails
with "Deployment failed: Not enough nodes - available: 0, requested:
2".

E.g. `ironic node-show compute-8` indicates that it is available and
properties/capabilities includes profile:compute.

What needs to be done to scale out using these settings to include the
remaining 8 compute nodes?

regards,


Gerard

-- 
Gerard Braad — 吉拉德
   F/OSS & IT Consultant in Beijing
   http://gbraad.nl  gpg: 0x592CFE75