Bay status is still CREATE_IN_PROGRESS after stack create failed

[FenghuaFang]

When a bay is failed create, i found the bay status is still CREATE_IN_PROGRESS .

[stack@localhost magnum(keystone_admin)]$ heat stack-list
+--------------------------------------+----------------------+---------------+----------------------+
| id | stack_name | stack_status | creation_time |
+--------------------------------------+----------------------+---------------+----------------------+
| 6ad0b1f4-ccc9-4155-b63d-9148f6e7ba7e | testbay-qu76axva3bls | CREATE_FAILED | 2015-04-15T08:07:59Z |
+--------------------------------------+----------------------+---------------+----------------------+
[stack@localhost magnum(keystone_admin)]$ magnum bay-list
+--------------------------------------+---------+------------+--------------------+
| uuid | name | node_count | status |
+--------------------------------------+---------+------------+--------------------+
| 363a48d6-4e44-46c8-ac9e-2b24bc6abcf7 | testbay | 2 | CREATE_IN_PROGRESS |
+--------------------------------------+---------+------------+--------------------+

This is the error messages on the magnum-conductor screen.

"
2015-04-15 04:41:21.873 ERROR magnum.openstack.common.loopingcall [-] in fixed duration looping call
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall Traceback (most recent call last):
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall File "/opt/stack/magnum/magnum/openstack/common/loopingcall.py", line 81, in _inner
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall self.f(*self.args, **self.kw)
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall File "/opt/stack/magnum/magnum/conductor/handlers/bay_k8s_heat.py", line 220, in poll_and_check
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall 'status': stack.stack_status})
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall KeyError: u'attempts'
2015-04-15 04:41:21.873 TRACE magnum.openstack.common.loopingcall
"

So, there is two advices
1.
Whether it's better to update the status of the by to "CREATE_FAILED"?
2.
The bay create progress stuck for a very long time in the status "CREATE_IN_PROGRESS".
Whether it's better to tell the user what magnum is doing now?

[Janek Lehr (jjlehr)]

I've seen this as well, but if the --timeout option is used during bay create then the bay shows CREATE_FAILED as it should. It looks like when max_attempts is exceeded the bay status is not updated.

[Adrian Otto (aotto)]

Please indicate steps to reproduce, and include logs of your attempts to help us make this bug actionable.

[Tom Cammann (tom-cammann)]

It would good to provide the heat logs for this failure also. However Janek's suggestion seems very likely and we need to decide what to do when max_attempts is exceeded or heat falls over. If heat has died in the background or such, there isn't going accurate we can report.

[Martin Falatic (martinfalatic)]

I am not able to reproduce this issue thus far. If you could provide details of what led to the issue, whether you found a workaround whether the issue is still affecting you and any logs or configuration information, that'd help me debug this.

[Eli Qiao (taget-9)]

I think this can be closed after periodic task bp finished https://blueprints.launchpad.net/magnum/+spec/add-periodic-task

[Eli Qiao (taget-9)]

patch is here.
https://review.openstack.org/#/c/194514/

[Martin Falatic (martinfalatic)]

Thank you for the update!