instances cannot be deleted if missing instance_info elements

[Adam Gandelman]

After migrating a running instance from nova-bm -> ironic, the instance cannot be deleted in nova. Doing so results in the following:

er': u'ironic', u'request_id': u'req-50ba4550-4df7-47dc-b34e-248856b24237', u'is_public_api': False, u'domain_id': u'default', u'tenant': u'service'} _safe_log /usr/local/lib/python2.7/dist-packages/oslo/messaging/_drivers/common.py:177
2014-09-12 22:32:47.751 9395 DEBUG ironic.conductor.manager [-] RPC do_node_tear_down called for node 2700db7f-2b15-49b8-bb3a-2b29dae28a20. do_node_tear_down /usr/local/lib/python2.7/dist-packages/ironic/conductor/manager.py:537
2014-09-12 22:32:47.751 9395 DEBUG ironic.conductor.task_manager [-] Attempting to reserve node 2700db7f-2b15-49b8-bb3a-2b29dae28a20 reserve_node /usr/local/lib/python2.7/dist-packages/ironic/conductor/task_manager.py:179
2014-09-12 22:32:53.216 9395 DEBUG oslo.messaging._drivers.amqp [-] UNIQUE_ID is 223db7da5a20428680fcfd59bd13f3b1. _add_unique_id /usr/local/lib/python2.7/dist-packages/oslo/messaging/_drivers/amqp.py:246
2014-09-12 22:32:53.217 9395 WARNING ironic.conductor.manager [-] Error in tear_down of node 2700db7f-2b15-49b8-bb3a-2b29dae28a20: Cannot validate iSCSI deploy. The following parameters were not passed to ironic: ['root_gb', 'image_source']
2014-09-12 22:32:53.221 9395 DEBUG oslo.messaging._drivers.amqp [-] UNIQUE_ID is 28b21ff6d18e4313aa89c8d2bdf37718. _add_unique_id /usr/local/lib/python2.7/dist-packages/oslo/messaging/_drivers/amqp.py:246
Traceback (most recent call last):
  File "/usr/local/lib/python2.7/dist-packages/eventlet/hubs/hub.py", line 455, in fire_timers
    timer()
  File "/usr/local/lib/python2.7/dist-packages/eventlet/hubs/timer.py", line 58, in __call__
    cb(*args, **kw)
  File "/usr/local/lib/python2.7/dist-packages/eventlet/greenthread.py", line 212, in main
    result = function(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/ironic/conductor/manager.py", line 593, in _do_node_tear_down
    node.target_provision_state = states.NOSTATE
  File "/usr/local/lib/python2.7/dist-packages/oslo/utils/excutils.py", line 82, in __exit__
    six.reraise(self.type_, self.value, self.tb)
  File "/usr/local/lib/python2.7/dist-packages/ironic/conductor/manager.py", line 584, in _do_node_tear_down
    task.driver.deploy.clean_up(task)
  File "/usr/local/lib/python2.7/dist-packages/ironic/drivers/modules/pxe.py", line 399, in clean_up
    pxe_info = _get_image_info(node, task.context)
  File "/usr/local/lib/python2.7/dist-packages/ironic/drivers/modules/pxe.py", line 223, in _get_image_info
    d_info = _parse_deploy_info(node)
  File "/usr/local/lib/python2.7/dist-packages/ironic/drivers/modules/pxe.py", line 140, in _parse_deploy_info
    info.update(iscsi_deploy.parse_instance_info(node))
  File "/usr/local/lib/python2.7/dist-packages/ironic/drivers/modules/iscsi_deploy.py", line 120, in parse_instance_info
    deploy_utils.check_for_missing_params(i_info, error_msg)
  File "/usr/local/lib/python2.7/dist-packages/ironic/drivers/modules/deploy_utils.py", line 408, in check_for_missing_params
    {'error_msg': error_msg, 'missing_info': missing_info})
MissingParameterValue: Cannot validate iSCSI deploy. The following parameters were not passed to ironic: ['root_gb', 'image_source']

[Adam Gandelman (gandelman-a)]

Two issues here caused by missing or incorrect instance_info data after running the ironic-nova-bm-migrate script:

* nova-bm uses root_mb, ironic uses root_gb but there is no conversion happening. Same issue /w ephemeral_mb vs ephemeral_gb, though that wouldn't cause issues here.

* 'image_source' is missing entirely from the node's instance info, it was never migrated to being with. The nova-bm table contains no reference to the image associated with booted instances. To migrate this from nova along with the other data, the migration script would need to also query the nova database directly

[aeva black (tenbrae)]

Updated bug title to reflect duplicate "node-delete operation on failed node does not work"

tldr; the PXE driver can not clean_up() if some parameters are missing from node.instance_info, leading to a failure in Nova. This situation can be the result of a failed launch or of a migration from nova "baremetal".

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/121615

[Adam Gandelman (gandelman-a)]

This also affects nodes with missing elements in the driver_info, specifically pxe_deploy_kernel + pxe_deploy_ramdisk

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/121615
Committed: https://git.openstack.org/cgit/openstack/ironic/commit/?id=985eba60edc339a1444e9206552eedb0e28758b6
Submitter: Jenkins
Branch: master

commit 985eba60edc339a1444e9206552eedb0e28758b6
Author: Jim Rollenhagen <email address hidden>
Date: Mon Sep 15 09:16:20 2014 -0700

    Allow clean_up with missing image ref

    This change allows clean_up to continue with missing image_source
    in Node.instance_info. This previouly failed in cases where:

    * A node failed to deploy and so didn't have image_source
    * A node was migrated from nova-baremetal

    Change-Id: I03543adc74f9da83fff58e0cffe34f36a055e19a
    Closes-Bug: 1368984