resize fail didn't show a correct info when --poll specified when resize down

[jichenjc]

jichen@controller:~$ nova resize --poll t9 12
Server resizing... 100% complete
Finished

see following info in nova compute.log and the resize failed, we should give accurate info about it

2014-05-28 15:29:12.508 4259 TRACE oslo.messaging.rpc.dispatcher self.instance_events.clear_events_for_instance(instance)
2014-05-28 15:29:12.508 4259 TRACE oslo.messaging.rpc.dispatcher File "/usr/lib/python2.7/contextlib.py", line 35, in __exit__
2014-05-28 15:29:12.508 4259 TRACE oslo.messaging.rpc.dispatcher self.gen.throw(type, value, traceback)
2014-05-28 15:29:12.508 4259 TRACE oslo.messaging.rpc.dispatcher File "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 5556, in _error_out_instance_on_exception
2014-05-28 15:29:12.508 4259 TRACE oslo.messaging.rpc.dispatcher raise error.inner_exception
2014-05-28 15:29:12.508 4259 TRACE oslo.messaging.rpc.dispatcher ResizeError: Resize error: Unable to resize disk down.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/101435

[melanie witt (melwitt)]

I'm leaning toward thinking this a nova bug as I learned more about how resize is supposed to work.

From this user documentation [0], it says the user issues the resize command to resize the instance up or down, if that was successful, the instance gets state VERIFY_RESIZE and the user is supposed to confirm or revert the resize to get the state to ACTIVE or ERROR. It seems like there should be another state like FAILED_RESIZE or something that indicates to the user that something went wrong. Setting state to ERROR would be too generic I think in this case, as it would imply, at least to me, that the instance is totally broken. Right now a failed resize will show status ACTIVE. I think if you know all the states and their transitions, that's a fine status, but if you don't, it's misleading and gives the idea that the resize worked when it didn't.

[0] http://docs.openstack.org/user-guide/content/nova_cli_resize.html

[jichenjc (jichenjc)]

ok, let me think about it , I will consider the alternatives you suggested and will continue to work on it
will be on vacation for a few days and will come back to you after that,

really appreciate your kindly comments and help ~ thanks a lot

[melanie witt (melwitt)]

@jichenjc Okay -- I'd like to hear others' thoughts too. Thanks for looking into it.

[jichenjc (jichenjc)]

@melanie, thanks for your comments it's not hard to add a state FAILED_RESIZE and I have following concerns

In the report of the bug ,if user want to resize with a smaller disk, it's kind of 'wrong' action, and from nova
perspective, it will do nothing but do some valid check, then nova decide not to do that action
so the instance is not changed at all ,only the 'action' failed but the instance is still ACTIVE
so we can take this defect into 2 steps , first is to notify user the 'resize' failed instead of silent
then we can consider whether we need to add FAILED_RESIZE feature since we might need a lot of other codes to handle this kind of state, e.g. in _init_instance, and I know we are going to have a 'task' state which will report those kind of failure instead of setting to ERROR state, maybe it will handle this better

thoughts? thanks a lot

[OpenStack Infra (hudson-openstack)]

Change abandoned by Kevin L. Mitchell (<email address hidden>) on branch: master
Review: https://review.openstack.org/101435
Reason: This change was last updated in November 2014, at which time it failed tests; also, reviewer comments have not been acted upon or responded to. Assuming this is abandoned…

[Sujitha (sujitha-neti)]

There are no open reviews for this bug report since a long time.
To signal that to other contributors which might provide patches for
this bug, I change the status from "In Progress" to "Confirmed" and
remove the assignee.
Feel free to add yourself as assignee if you plan to work on it.

[Markus Zoeller (markus_z) (mzoeller)]

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 (LIBERTY, MITAKA, OCATA, NEWTON).
  Valid example: CONFIRMED FOR: LIBERTY

[Takashi Natsume (natsume-takashi)]

CONFIRMED FOR: master (commit ff47787e11a0a1b6299298b25b4c982bcfae6e1c)

[Takashi Natsume (natsume-takashi)]

CONFIRMED FOR: nova master (commit ff47787e11a0a1b6299298b25b4c982bcfae6e1c)
CONFIRMED FOR: python-novaclient 10.1.0