[pike] Openstack overcloud deploy failed: not enough nodes

[Cédric Jeanneret deactivated]

Dear Stackers,

I have a weird situation:

- `openstack overcloud profiles list` returns the intended list of node <-> profiles matching as shown here: http://paste.openstack.org/show/620290/
- `openstack overcloud profiles match --control-flavor xx --compute-flavor yy --control-scale 1 --compute-scale 2` returns 0 as show here: http://paste.openstack.org/show/620291/
- nodes are in "available" mode, and "maintenance" is set to False as shown here: http://paste.openstack.org/show/620292/

But, unfortunately, for some reason I'm unable to find/understand, I get the following error while trying to deploy the overcloud:
http://paste.openstack.org/show/620293/

I really don't understand what's going on :(. And I have no idea where to start the search - as you can see, I already ensured all should work as expected…

Thank you!

Cheers,

C.

[Steven Hardy (shardy)]

Please can you provide the full deploy command used, and the output of openstack baremetal node show for each node, and the openstack flavor show output for each flavor used in the deploy command, thanks!

[Cédric Jeanneret deactivated (cjeanneret-c2c-deactivated)]

Hello Steven,

The full command is as follow:
openstack --verbose overcloud deploy --templates ./openstack-tripleo-heat-templates -e ./environment.yaml -e ./openstack-tripleo-heat-templates/environments/network-isolation.yaml -e ./openstack-tripleo-heat-templates/environments/net-bond-with-vlans.yaml -e ./openstack-tripleo-heat-templates/environments/tls-endpoints-public-dns.yaml

Note: openstack-tripleo-heat-templates is a local directory with the content of /usr/share/openstack-tripleo-heat-templates copied in it so that we can follow the changes in our git.

The environment.yaml is a local yaml file containing some parameter_defaults entries (for network, and so on).

Here are the requested outputs:
openstack baremetal node show *: http://paste.openstack.org/show/620300/
openstack flavor show *: http://paste.openstack.org/show/620301/

The requested information from environment.yaml (*Flavor): http://paste.openstack.org/show/620302/

Hope that helps.

Cheers,

C.

[Cédric Jeanneret deactivated (cjeanneret-c2c-deactivated)]

As requested by "dtantsur" on #openstack-ironic, here's the nova-compute.log

Apparently, the issue is somewhere between nova which doesn't see the nodes, and ironic, which actually see them properly.

We might be able to say "it's probably not a profile mismatch issue".

[Cédric Jeanneret deactivated (cjeanneret-c2c-deactivated)]

Apparently, it's a duplicate of #1513068 - have to confirm the patch resolves the issue (openstack undercloud upgrade is running with a corrected template, according to https://github.com/openstack/instack-undercloud/commit/b366467f6d596017913167cf3e6c158805c8ae52 )

Please wait for my feedback before marking this issue as duplicated.

[Cédric Jeanneret deactivated (cjeanneret-c2c-deactivated)]

Me again,

Thanks to Dmitry (dtantsur) I could validate my issue is a duplicate of #1513068 - the resolution was:
- apply the patch linked in my previous comment
- `openstack undercloud upgrade'
- `openstack compute service set --enable <COMPUTE HOST> nova-compute'

Thanks to all of the people on #tripleo and #openstack-ironic for their help :).

Cheers,

C.