ResourceFailure: in heat returns "unknown" error

[Cédric Jeanneret deactivated]

Dear Stackers,

While trying to add ceph nodes to my current tripleO deploy on Pike, I have the following issue in heat logs:
http://paste.openstack.org/show/623181/

This issue is reported to mistral as follow:
http://paste.openstack.org/show/623182/

I don't really know what to provide in addition, feel free to ask me.

The nodes were added using the following commands:
openstack overcloud node import baremetal.json
openstack overcloud node introspect --all-manageable --provide

one of the node is described as follow with `openstack baremetal node show ceph-storage-0'
http://paste.openstack.org/show/623183/

Deploy command includes the following specific configuration:
http://paste.openstack.org/show/623184/

Thank you!

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

I found something more in nova-conductor log:
http://paste.openstack.org/show/623186/

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

After some more digging, in nova-compute.log:
http://paste.openstack.org/show/623187/

Apparently, novajoin-server issue, although I didn't activate the novajoin thingy with the server metadata o_O

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

After more searches, it appears it's a disk space issue on the undercloud:
InstanceDeployFailure: Failed to provision instance 0cf68149-bae6-432c-93b2-6d58d2a48661: Failed to deploy. Error: Disk volume where '/var/lib/ironic/master_images/tmpjEgvzp' is located doesn't have enough disk space. Required 5211 MiB, only 4467 MiB available space present.

The undercloud we're running on had about 4.9G of free space, this matches the amount in the log. This log is from nova-compute, btw.

It would be good to raise that one into mistral so that it actually hit the user in the eye. Another idea would be to get a pre-check flight for that. Indeed, this issue is raised long, long after the start, when the deploy process actually starts the node provisionning: ironic wants to fetch the image from glance storage, and can't do so because it lacks the necessary space.

Cheers,

C.

[Luke Short (ekultails)]

I believe this may now be addressed by this validation: https://github.com/openstack/tripleo-validations/tree/master/roles/undercloud-disk-space