| opendevreview | Yian Zong proposed openstack/manila master: Dell PowerStore and PowerFlex Manila drivers enhancement https://review.opendev.org/c/openstack/manila/+/893322 | 04:30 |
|---|---|---|
| dviroel | felipe_rodrigues: it seems that your CI is broken | 17:33 |
| dviroel | felipe_rodrigues: "nova.exception.InternalError: Nova requires libvirt version 7.0.0 or greater" | 17:34 |
| dviroel | felipe_rodrigues: you would need to move to a newer version of Ubuntu | 17:37 |
| opendevreview | Ashley Rodriguez proposed openstack/manila master: Metadata for Share Export Location https://review.opendev.org/c/openstack/manila/+/887543 | 17:41 |
| opendevreview | Ashley Rodriguez proposed openstack/python-manilaclient master: Metadata for share export location https://review.opendev.org/c/openstack/python-manilaclient/+/889677 | 17:41 |
| felipe_rodrigues | thank you dviroel.. I triggered again just to have a second attempt.. | 17:46 |
| felipe_rodrigues | actually, if we see the execution before this one that you mentioned [1], the CI "passed".. It failed only one single test in the no-dhss job "test_soft_delete_and_restore_share" . This test is known as unstable, even the master branch fails sometimes with it. We are investigating why, it is a bit hard because it is not easy to reproduce. My point is: the patch could be seen as "passed" on our CI. We are | 17:54 |
| felipe_rodrigues | confident about it. | 17:54 |
| felipe_rodrigues | [1] https://review.opendev.org/c/openstack/manila/+/890291/5#message-8a11966f4bcbe58c27c96c2d07b20a2047e9fcb3 | 17:54 |
| felipe_rodrigues | I would like to hear about you guys.. If we have this problem with the upgrade the base image, we cannot fix today. It would take at least 1 or 2 days for fixing the CI. | 17:56 |
| dviroel | felipe_rodrigues: awesome that you guys set up an active iq job | 18:04 |
| dviroel | felipe_rodrigues: is the following error expected? | 18:05 |
| dviroel | RESPONSE: 500 Body=b'{"error":{"message":"NoneGB","target":null,"code":null}}' | 18:05 |
| dviroel | felipe_rodrigues: just a few of them, maybe negative tests? | 18:07 |
| dviroel | felipe_rodrigues: if you know that is working from your side and community ci is passing, I don't see as a blocker. As you mentioned, there are logs that show the weigher working | 18:11 |
| dviroel | felipe_rodrigues: I will let the W+1 with carloss | 18:12 |
| carloss | ++ - I'm okay pulling the trigger | 18:13 |
| carloss | was just watching the CI outputs to see if we could get some clean runs | 18:13 |
| felipe_rodrigues | Great catch dviroel! You should visit Manila more times haha | 18:52 |
| felipe_rodrigues | Let me explain: the ActiveIQ weight request requires the size of the object. The Manila group scheudling process just take in account the type (which makes sense, because each element can have a different size). I mean, the selection of one of the hosts for a new group doesn't have the "size" attribute. As result, it cannot run with ActiveIQ. | 18:58 |
| felipe_rodrigues | The error is not affecting the results, because when the API fails, the weigher is skipped (as it should be for scheduling groups). However, it is consuming the API unnecessarily. | 19:00 |
| felipe_rodrigues | We should have a check in the beginning of it: if it is weighing a share creation proceed, otherwise, skip it. The idea of this weigher is only for share creation (with size as part of the request). | 19:02 |
| dviroel | felipe_rodrigues: yeah, there are lots of api calls there, which I think that needs an optimization for better user exp | 19:05 |
| felipe_rodrigues | For sure, thanks for the feedback, it seems very important. | 19:05 |
| felipe_rodrigues | Confirmed: Manila NetApp CI is broken. The second execution had the same issue deploying the env. | 19:11 |
| felipe_rodrigues | https://review.opendev.org/c/openstack/manila/+/890291/5#message-eea5269c610e7887d6d63791f19c535cc836e4ac | 19:12 |
| felipe_rodrigues | We have fews problems to solve: | 19:18 |
| felipe_rodrigues | 1. Fix small points with the release note | 19:18 |
| felipe_rodrigues | 2. Improve the coverage for some corner cases (as pointed by carloss) | 19:18 |
| felipe_rodrigues | 3. Optimize the weigher, skipping for non share creation flows (as found by dviroel) | 19:18 |
| felipe_rodrigues | For those modifications, we should run the CI again. However, it is broken, given the Nova modifications. We cannot fix it today, it requires few days (1 or 2). | 19:18 |
| felipe_rodrigues | How should we proceed ? I am thinking about two alternatives. | 19:18 |
| felipe_rodrigues | 1. Extend the freeze until next friday (FFE) | 19:18 |
| felipe_rodrigues | 2. Merge and then work on a follow-up patch | 19:18 |
| felipe_rodrigues | Wdyt carloss, dviroel and others ? | 19:19 |
| carloss | 2 it is :) | 20:00 |
| carloss | felipe_rodrigues: please address the comments and get some CI runs as soon as you can | 20:00 |
| felipe_rodrigues | thks carloss | 20:04 |
| carloss | anytime felipe_rodrigues | 20:35 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!