Erroneously determined to be completed when updating a container

[Ayumu Ueha]

The following error is occured in test_container_update_multi_kinds method of multinode-sol job.

```
ft1.1: tacker.tests.functional.sol_kubernetes.vnflcm.test_kubernetes_container_update.VnfLcmKubernetesContainerUpdate.test_container_update_multi_kindstesttools.testresult.real._StringException: Traceback (most recent call last):
  File "/home/zuul/src/opendev.org/openstack/tacker/tacker/tests/functional/sol_kubernetes/vnflcm/test_kubernetes_container_update.py", line 98, in test_container_update_multi_kinds
    self.assertNotEqual(before_resource['resourceId'],
  File "/usr/lib/python3.8/unittest/case.py", line 921, in assertNotEqual
    raise self.failureException(msg)
AssertionError: 'vdu1-update-5b9d95d894-l8xnp' == 'vdu1-update-5b9d95d894-l8xnp'
```

In my investigation, this is due to the fact that the update process has not completed yet, but lcm process has been finished.

[Before replace api]
  vdu1-update-764bbf8846-q5hzh: phase=RUNNING

[After replace api]
  vdu1-update-764bbf8846-q5hzh: phase=RUNNING
  vdu1-update-5b9d95d894-l8xnp: phase=PENDING <- newly created

[Erroneously determined state]
  vdu1-update-764bbf8846-q5hzh: phase=RUNNING <- not yet deleted
  vdu1-update-5b9d95d894-l8xnp: phase=RUNNING <- become to "RUNNING"

  ==> Judge as waiting completed <- wrong!!

[Ideal state of judging as finished]
  vdu1-update-5b9d95d894-l8xnp: phase=RUNNING

  ==> Old pod is deleted completely.
      (number of associated Pods is the same as Replicas.)

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/tacker/+/866583

[Yasufumi Ogawa (yasufum)]

Let me confirm the cause of the failure because your description is something ambiguous. Is the point of the problem just an old process is still running although it's expected to be terminated? So, you mentioned "[Erroneously determined state]" is not erroneously actually because the status RUNNING is not wrong actually, correct?

[Ayumu Ueha (ueha)]

Thank you for your question.

> Is the point of the problem just an old process is still running although it's expected to be terminated?

No. The problem is that the old Pod that is being replaced by replace remains and has stopped waiting for the replace to complete.

> So, you mentioned "[Erroneously determined state]" is not erroneously actually because the status RUNNING is not wrong actually, correct?

Kubernetes also returns the state of the pod as "Running" while it is being removed.
In the current implementation, when checking to see if it is really removed and replace is complete, it will decide that it is OK to leave the old Pods that are being removed.

This really need to exit the wait process when the completely old pod is removed.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/openstack/tacker/+/866583
Committed: https://opendev.org/openstack/tacker/commit/be6d22b40986f0d0030c02cce43057aa8fa2a281
Submitter: "Zuul (22348)"
Branch: master

commit be6d22b40986f0d0030c02cce43057aa8fa2a281
Author: Ayumu Ueha <email address hidden>
Date: Mon Dec 5 14:12:44 2022 +0000

    Fix waiting for pod removal when container update

    This patch fixes the problem of aborting the wait process even if there
    are pods left to be removed by replace when container update feature.

    Closes-Bug: #1998756
    Change-Id: I6bc5c10fe4c7f646b84798d3b4a296721ccea0f8

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/tacker 9.0.0.0rc1 release candidate.