| *** jhesketh has quit IRC | 01:53 | |
| *** jhesketh has joined #refstack | 02:22 | |
| *** jhesketh has quit IRC | 03:29 | |
| *** jhesketh has joined #refstack | 03:36 | |
| openstackgerrit | Martin Kopec proposed openstack/python-tempestconf master: Add Queens and Rocky job https://review.opendev.org/675105 | 05:09 |
|---|---|---|
| *** tosky has joined #refstack | 07:17 | |
| *** kopecmartin|off is now known as kopecmartin | 07:59 | |
| kopecmartin | tosky: arxcruz can someone explain this? if i use: pip install -c constrains <path> , the project is not installed correctly, if after that i try: pip install <path> , the packages are updated based on the requirements.txt and constrains is not taken into account (expected) - BUT now the project is installed correctly | 08:32 |
| kopecmartin | http://paste.openstack.org/show/755644/ | 08:32 |
| kopecmartin | by correctly i mean that i can run discover-tempest-config command without it complaining about NoneType error on an internal method | 08:32 |
| tosky | kopecmartin: is this tempestconf with the os-client-config->openstacksdk patch, or the current master? | 08:33 |
| kopecmartin | with that patch | 08:33 |
| kopecmartin | or the current master as well, let me try | 08:33 |
| kopecmartin | i thing it's not a problem of the patch | 08:34 |
| kopecmartin | OMG *think :D i need another coffee | 08:36 |
| kopecmartin | shit, maybe it it connected to the patch :o | 08:36 |
| *** lpiwowar has joined #refstack | 08:38 | |
| tosky | "ERROR: python-tempestconf 2.2.2.dev7 has requirement openstacksdk>=0.32.0, but you'll have openstacksdk 0.27.0 which is incompatible." | 08:38 |
| tosky | I was almost sure it was connected | 08:38 |
| kopecmartin | i know about the error, however, pip doesn't fail on it and the openstacksdk 0.27 version is installed | 08:39 |
| kopecmartin | if that's the problem then fine, we know that our new job will work | 08:39 |
| kopecmartin | but i'm not sure what exactly happened there | 08:39 |
| tosky | maybe it makes sense to disconnect the jobs patch from the openstacksdk patch, and merge the former first | 08:43 |
| openstackgerrit | Martin Kopec proposed openstack/python-tempestconf master: Add Queens and Rocky job https://review.opendev.org/675105 | 08:50 |
| kopecmartin | yeah, let's do that | 08:50 |
| openstackgerrit | Lukas Piwowarski proposed openstack/python-tempestconf master: Replace os-client-config module by openstacksdk https://review.opendev.org/674635 | 10:16 |
| kopecmartin | tosky: seems it's related to the patch, but i have no idea why yet https://logs.opendev.org/35/674635/5/check/python-tempestconf-tempest-devstack-admin-rocky/6ef8698/job-output.txt.gz#_2019-08-08_10_59_37_471241 | 11:41 |
| kopecmartin | correct constraints is used and there is no error during deps installation https://logs.opendev.org/35/674635/5/check/python-tempestconf-tempest-devstack-admin-rocky/6ef8698/job-output.txt.gz#_2019-08-08_10_59_06_809805 | 11:42 |
| kopecmartin | interesting, maybe import issue? | 11:42 |
| kopecmartin | tosky: lpiwowar just made a good point, there might be a problem with openstacksdk in the version for queens and rocky :o | 11:49 |
| tosky | kopecmartin: that was the meaning of my original content: openstacksdk changed its interface over time | 12:24 |
| tosky | s/content/comment/ | 12:24 |
| openstackgerrit | Lukas Piwowarski proposed openstack/python-tempestconf master: Replace os-client-config module by openstacksdk https://review.opendev.org/674635 | 12:40 |
| *** lpiwowar has quit IRC | 12:40 | |
| arxcruz | tosky: kopecmartin so, is the quees/rocky job will be an experimental one or normal job ? | 13:06 |
| kopecmartin | arxcruz: i'd say normal, seems that we have to keep an eye on openstacksdk | 13:09 |
| arxcruz | ok | 13:09 |
| kopecmartin | it doesn't require a lot of resources, does it? | 13:10 |
| tosky | shouldn't we add a stein one too? | 13:11 |
| tosky | maybe we can trim down the resources by adding just the -admin jobs for the older versions to the check and gate queue , and add their -demo counterparts to the experimental queue | 13:12 |
| tosky | non sure whether the question about "experimental jobs" has been addressed, but experimental jobs is just a shortcut for "jobs defined in the experimental queue which are executed by adding a 'check experimental' comment" | 13:13 |
| kopecmartin | ahaa, in that case, we can make all the new jobs expoerimental ones .. we'll check just from time to time if they're working .. and mainly when we will be releasing a new version of tempestconf | 13:14 |
| arxcruz | tosky: thanks, yeah, i did not explain, i was about to do this morning, but, well, my brain doesn't work before caffeine :D | 13:15 |
| tosky | we shouldn't forget | 13:15 |
| tosky | we could make them also (not just, but also) periodic | 13:15 |
| tosky | I'm not sure how it works with notifications, but it should be easy to figure out | 13:15 |
| arxcruz | periodic you don't get notification because it's not running as a change on gerrit | 13:26 |
| arxcruz | and notifications came from gerrit | 13:26 |
| arxcruz | however, you can use openstack-health and subscribe to the rss feed | 13:26 |
| arxcruz | or we can add this job on cokpick on tripleo-ci | 13:27 |
| tosky | I see | 13:28 |
| tosky | so what do you think? Right now we would need 6 new devstack jobs | 13:30 |
| kopecmartin | tosky: when adding jobs for older releases , we're talking about check jobs, right? or they have to be also gate ones? | 13:44 |
| tosky | kopecmartin: also gate | 13:44 |
| kopecmartin | ok | 13:44 |
| tosky | while we don't share much queues with other jobs, it may be possible that we could land two patches and the order would matter | 13:45 |
| tosky | as we commit to support the older versions, this commitment should be really strong, and the jobs are the way to show it | 13:45 |
| tosky | that's why I proposed to have at least one job for each older branches in the check and gate queues | 13:46 |
| kopecmartin | tosky: good, makes sense | 13:47 |
| tosky | but two may be too many for the available resources; but of course I will buy any other idea that allows us to not regress in the older branches without killing -infra :) | 13:47 |
| kopecmartin | we don't have that big patch cadence, so it should be fien | 13:48 |
| kopecmartin | fine | 13:48 |
| openstackgerrit | Martin Kopec proposed openstack/python-tempestconf master: Add Queens, Rocky and Stein job https://review.opendev.org/675105 | 13:50 |
| arxcruz | I though we have releases of tempestconf for each version... | 13:57 |
| tosky | no, tempestconf is branchless | 13:57 |
| tosky | then it's up to us to decide whether we want to ask the consumers (like RDO) to adopt the new version for older branches | 13:57 |
| arxcruz | ack | 13:58 |
| tosky | but not testing it beforehand it's calling for troubles | 13:58 |
| *** tosky has quit IRC | 16:49 | |
| *** kopecmartin is now known as kopecmartin|off | 17:31 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!