Missing nodes available to assign

[Martin Mágr]

Description of problem:
After deployment 1 controller and 1 compute via tripleo-quickstart, I have 4 more nodes available to deploy on (see attached screenshots). However, I'm able to deploy only to two nodes.

Additional info:
According to validation scripts my stack was in pretty crappy state, so maybe that's a reason.

[Martin Mágr (mmagr)]

[stack@undercloud ~]$ ironic node-list
+--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
| UUID | Name | Instance UUID | Power State | Provisioning State | Maintenance |
+--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
| 73a91814-0767-4b9e-bb57-aea8b9521c63 | control-0 | 2eeecde7-a96f-4186-a680-27f2c8c1f59b | power on | active | False |
| fe7652d4-f604-4d2c-ad1a-343a7abe8dea | control-1 | None | power off | available | False |
| 9a318d00-ccf8-465d-9be6-42b4c8262c7f | control-2 | None | power off | available | False |
| 3cc3e2bc-f9e5-4ea7-9493-6bcbf06b246a | compute-0 | 47fe172e-27de-426f-8cbf-dddc991dac60 | power on | active | False |
| 1243f0de-7bf2-4f46-bac0-5853a8af4997 | ceph-0 | None | power off | available | False |
| 57975a73-5180-4c99-b8e3-736d2d23580c | swift-0 | None | power off | available | False |
+--------------------------------------+-----------+--------------------------------------+-------------+--------------------+-------------+
[stack@undercloud ~]$ ironic node-show fe7652d4-f604-4d2c-ad1a-343a7abe8dea
+------------------------+--------------------------------------------------------------------------+
| Property | Value |
+------------------------+--------------------------------------------------------------------------+
| chassis_uuid | |
| clean_step | {} |
| console_enabled | False |
| created_at | 2016-11-28T13:39:43+00:00 |
| driver | pxe_ssh |
| driver_info | {u'ssh_username': u'stack', u'deploy_kernel': u'63c8157c- |
| | 80ec-4314-b4a0-275964dfdc44', u'deploy_ramdisk': u'b86f2166-6580-4467 |
| | -a6fc-571a5fdadf27', u'ssh_key_contents': u'-----BEGIN RSA PRIVATE KEY-- |
| | --- |
| | MIIJKgIBAAKCAgEAx2YIElWQdU7LZQcPExeSOpvCfNraUfdYyI4MvOz5Dv6AogeM |
| | 5 |
| | Fv/3MvPtlaWNtdCZ7s3RTGj6z48tswqdPqZ+ipJDInojNia2A3qPfCl5+kublku |
| | lHcXuXw |
| | QD/AMdA+vmIActIs...

[Martin Mágr (mmagr)]

Forgot to mention, that I skipped node introspection.

[Julie Pichon (jpichon)]

Ah, I just filed a duplicate (bug 1647648). Thanks for filing this bug! I'll copy the information here and close the dup. This is due to incompatibility with how Quickstart creates flavors for profile matching. You can see it in the Ironic input provided: 'profile:objectstorage", "profile:ceph".

[Julie Pichon (jpichon)]

Also we use the tripleo tracker for our bugs, I'll move that one over there :)

[Julie Pichon (jpichon)]

The flavors set up by Quickstart for Ceph and Object Storage are different than the ones the UI (and perhaps CLI?) use and expect for profile matching.

See http://docs.openstack.org/developer/tripleo-quickstart/configuration.html#setting-number-and-type-of-overcloud-nodes : Quickstart expects 'ceph', 'objectstorage'. The UI expects 'ceph-storage', 'swift-storage'. (I tried just setting 'ceph-storage' in the quickstart config file but the install then failed because of this, so it's not easy to work around.)

From the UI perspective it's pretty bad because of bug 1644801 (cannot unassign custom roles if they're not defined in your current plan), as the nodes are treated like if they were associated with a 'custom role', so they are not shown on any of the default role 'cards' in the current Deployment Plan. They still show as Available/Not assigned in the nodes, but because it's not possible to remove or change the role then, they remain inaccessible and unavailable for deployments.

[Martin Mágr (mmagr)]

eems like it's a problem of flavor/capabilities settings of quickstart:

[stack@undercloud ~]$ openstack server show -c name -c flavor 9e2fd56a-fb7e-4f23-be76-1d6739780710
+--------+-----------------------------------------------------------+
| Field | Value |
+--------+-----------------------------------------------------------+
| flavor | oooq_objectstorage (fad64184-1589-4d92-bc0b-ff08f032573f) |
| name | overcloud-objectstorage-0 |
+--------+-----------------------------------------------------------+
[stack@undercloud ~]$ openstack server show -c name -c flavor fc440a25-34db-4dad-a0f8-0f8f78e79f8b
+--------+--------------------------------------------------+
| Field | Value |
+--------+--------------------------------------------------+
| flavor | oooq_ceph (b79b1415-8d63-4204-9a94-1467ce7e5039) |
| name | overcloud-cephstorage-0 |
+--------+--------------------------------------------------+
[stack@undercloud ~]$

------

[stack@undercloud ~]$ ironic node-show 1243f0de-7bf2-4f46-bac0-5853a8af4997 | grep capabilities
| | u'capabilities': u'{"profile": "ceph", "boot_option": "local"}', |
| | u'cpus': u'1', u'capabilities': u'profile:ceph,boot_option:local'} |
[stack@undercloud ~]$ ironic node-show 57975a73-5180-4c99-b8e3-736d2d23580c | grep capabilities
| | u'capabilities': u'{"profile": "objectstorage", "boot_option": |
| | u'cpus': u'1', u'capabilities': |
[stack@undercloud ~]$

I'll let this bug open so that somebody from UI team can doublecheck this, but feel free to close it in case my guess is correct.

[Jiri Tomasek (jtomasek)]

IIUC the profile tag needs to match the flavor name and flavor's profile tag. So the one which GUI uses should be correct. I'd think that it is a quickstart bug. Correct me if I am wrong please

[Julie Pichon (jpichon)]

Right, the UI profile names match the Nova flavors:

[stack@undercloud ~]$ openstack flavor list
+--------------------------------------+---------------+------+------+-----------+-------+-----------+
| ID | Name | RAM | Disk | Ephemeral | VCPUs | Is Public |
+--------------------------------------+---------------+------+------+-----------+-------+-----------+
| 14d8f399-8f01-4325-84ed-b8cbc51028ea | ceph-storage | 4096 | 40 | 0 | 1 | True |
| 695ac08b-a8c4-490a-ac0f-8163dd7a1641 | compute | 4096 | 40 | 0 | 1 | True |
| 9e9d614b-693e-434a-90c7-90afce3532d2 | swift-storage | 4096 | 40 | 0 | 1 | True |
| a4dee923-e16a-480d-920c-511aea638eba | baremetal | 4096 | 40 | 0 | 1 | True |
| b795b324-c634-4907-97e4-2096d043ac87 | block-storage | 4096 | 40 | 0 | 1 | True |
| cde5a023-2eea-4c5f-9097-f96662a35ed8 | control | 4096 | 40 | 0 | 1 | True |
+--------------------------------------+---------------+------+------+-----------+-------+-----------+

I'll add quickstart to this bug as well and perhaps they can comment.

[Julie Pichon (jpichon)]

Somewhat related: bug 1647736.

[Jiri Tomasek (jtomasek)]

We need to re-triage this one and close it if it has been fixed on tripleo-quickstart side

[Emilien Macchi (emilienm)]

This is an automated cleanup. This bug report has been closed because it
is older than 18 months and there is no open code change to fix this.
After this time it is unlikely that the circumstances which lead to
the observed issue can be reproduced.

If you can reproduce the bug, please:
* reopen the bug report (set to status "New")
* AND add the detailed steps to reproduce the issue (if applicable)
* AND leave a comment "CONFIRMED FOR: <RELEASE_NAME>"
  Only still supported release names are valid (FUTURE, PIKE, QUEENS, ROCKY, STEIN).
  Valid example: CONFIRMED FOR: FUTURE