[Starlingx-discuss] [BUG] armada stuck at pod waiting due to no pod exist for chart

[Lin, Shuicheng]

Hi all,
When I debug LP issue [0], I suspect it is armada issue, so I created story [1] in armada project.
The issue is that
"
StarlingX uses armada to manage the helm chart. And we find a bug [0] recently. After debug it, it should be due to armada assume there is at least 1 pod for each chart in the wait for resource become ready logic. But it is not true for some corner case. Such as the osh-openstack-ceph-rgw chart in StarlingX. Currently, it has 3 job only which requires one time execution. The pod for job is cleared after host reboot, so when do chart re-apply, there is no pod exist for the chart. And 'required' is set to True for pod in default, which indicates there is at least 1 pod. This lead to function _watch_resource_completions in wait.py stuck at w.stream(self.get_resources, **kwargs).
"
Could you help review my comments, and share me your suggestion for the issue? And do you agree with my fix suggestion in the story [1]?
Thanks.


[0]: https://bugs.launchpad.net/starlingx/+bug/1833609
[1]: https://storyboard.openstack.org/#!/story/2006133

Best Regards
Shuicheng

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.starlingx.io/pipermail/starlingx-discuss/attachments/20190701/fd61025d/attachment.html>