| *** pojadhav|out is now known as pojadhav|ruck | 04:59 | |
| opendevreview | wangjiaqi proposed openstack/devstack master: remove unicode from code https://review.opendev.org/c/openstack/devstack/+/852318 | 05:37 |
|---|---|---|
| *** jpena|off is now known as jpena | 07:36 | |
| *** elodilles_pto is now known as elodilles | 08:06 | |
| *** brinzhang_ is now known as brinzhang | 09:09 | |
| *** soniya29 is now known as soniya29|afk | 10:26 | |
| *** soniya29|afk is now known as soniya29 | 11:31 | |
| opendevreview | Emma Foley proposed openstack/devstack stable/xena: Respect constraints on tempest venv consistently https://review.opendev.org/c/openstack/devstack/+/855146 | 14:41 |
| gmann | frickler: can you check this https://review.opendev.org/c/openstack/devstack/+/848482 | 14:46 |
| gmann | frickler: this is needed by ceilometer gate to unblock https://review.opendev.org/c/openstack/ceilometer/+/853668 | 14:51 |
| frickler | gmann: I must admit that I stopped trying to understand this tempest venv requirements voodoo. feel free to self-approve if you feel that this works | 14:52 |
| gmann | frickler: it worked, it is tested in https://review.opendev.org/c/openstack/ceilometer/+/853668 | 14:53 |
| frickler | #startmeeting qa | 15:01 |
| opendevmeet | Meeting started Tue Aug 30 15:01:44 2022 UTC and is due to finish in 60 minutes. The chair is frickler. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:01 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:01 |
| opendevmeet | The meeting name has been set to 'qa' | 15:01 |
| gmann | o/ | 15:01 |
| frickler | #topic open discussion | 15:01 |
| frickler | since martin is a away, no full meeting today, but option to discuss anything | 15:02 |
| * frickler can't type today | 15:03 | |
| soniya29 | hello | 15:03 |
| gmann | any info when martin will be back? | 15:04 |
| frickler | I think he said this meeting and next week he would miss | 15:05 |
| gmann | ok. i sent him the PTL nomination deadline over his personal email but not sure if he saw that as no response | 15:05 |
| frickler | 17:05 < kopecmartin> August 26th - September 9th | 15:05 |
| frickler | seems he might have just missed the announcement of the nomination period starting | 15:06 |
| gmann | yeah and may be many more other member too as this is vacation period. in TC we are discussing about extending the nomination but it is not final yet | 15:07 |
| frickler | worst case TC will have to assign him as PTL later, I guess | 15:08 |
| frickler | I don't expect anyone else will volunteer | 15:09 |
| gmann | yeah and this will be easy fill as election dates are little odd here | 15:09 |
| gmann | will make note about it in TC assignment if needed | 15:09 |
| frickler | regarding 17:05 < kopecmartin> August 26th - September 9th | 15:11 |
| frickler | oops, wrong paste buffer | 15:11 |
| frickler | regarding https://review.opendev.org/c/openstack/devstack/+/848482 I would like to leave that up to tempest cores | 15:11 |
| gmann | no one else then me to approve. or dansmith ^^ if you want to review it? | 15:12 |
| gmann | anyways, will wait for this afternoon | 15:13 |
| frickler | nothing wrong with single-approving IMO, in particular for gate fixes | 15:13 |
| dansmith | got it | 15:13 |
| gmann | sure | 15:13 |
| frickler | any other topic? | 15:15 |
| gmann | nothing else from me. | 15:15 |
| frickler | ah, maybe one thing, I added the lib/neutron removal to the PTG etherpad and also for Neutron and mentioned it in their meeting today | 15:16 |
| frickler | we have that planned to happen in the A cycle | 15:17 |
| gmann | +1, that will be good | 15:17 |
| gmann | thanks | 15:17 |
| frickler | and I proposed to do a joint session at the PTG, hoping maybe someone from the Neutron team will be doing most of the work ;) | 15:17 |
| frickler | but volunteers from the QA side would certainly be welcome, too | 15:18 |
| gmann | sounds good | 15:18 |
| soniya29 | +1 | 15:20 |
| frickler | I just went through all the jobs check lists, Fedora still seems unstable, also due to be switched to F36 soon. everything else looks normal | 15:22 |
| frickler | so I think that's it for today, thx everyone | 15:22 |
| frickler | #endmeeting | 15:23 |
| opendevmeet | Meeting ended Tue Aug 30 15:23:03 2022 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:23 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/qa/2022/qa.2022-08-30-15.01.html | 15:23 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/qa/2022/qa.2022-08-30-15.01.txt | 15:23 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/qa/2022/qa.2022-08-30-15.01.log.html | 15:23 |
| soniya29 | frickler, thanks | 15:23 |
| gmann | thanks frickler | 15:23 |
| *** artom__ is now known as artom | 15:34 | |
| dansmith | do we have a gate-blocking requirements thing that just happened? | 15:55 |
| dansmith | https://zuul.opendev.org/t/openstack/build/5088531548e2497f8a59ddd8c8b89de8 | 15:55 |
| dansmith | https://zuul.opendev.org/t/openstack/build/9935703405454ba292fa2cb2746377d7 | 15:55 |
| dansmith | https://zuul.opendev.org/t/openstack/build/576dded2b5814b75b7ab9588f439608d | 15:56 |
| frickler | at first sight that looks more like the usual pypi hickup we see occasionally | 15:57 |
| gmann | yeah, seems pypi index issue | 15:57 |
| gmann | jinja2 3.1.2 depends on MarkupSafe>=2.0 | 15:57 |
| gmann | The user requested (constraint) markupsafe===2.1.1 | 15:57 |
| gmann | constraints seems ok | 15:57 |
| dansmith | but that many in the last ten minutes? | 15:58 |
| frickler | if the main index is broken, likely all jobs will fail | 15:58 |
| dansmith | oh, pypi not pip, meaning something on pypi server side that will clear? | 15:58 |
| clarkb | its always good to double check they havne't deleted a package version (they didn't delete markupsafe 2.1.1) and that the python versions for the cosntrained package matches the local python this case is >=python3.7 | 15:58 |
| clarkb | But I agree this appears like pypi index problems | 15:59 |
| dansmith | ack okay | 15:59 |
| clarkb | dansmith: ya what happens is PyPIs CDN has two backends. The primary backend and the fallback backend. If the CDN needs to refresh data it tries the primary. If the primary fails it goes to the backup. The problem is the backup is usually significantly out of date | 15:59 |
| dansmith | the first two were docs related so I figured they were the same issue, but the packages are different | 15:59 |
| clarkb | This is particularly problematic for openstack due to the use of constraints that update fairly quickly to match newly released package versions | 16:00 |
| dansmith | actually I guess they're all docs | 16:00 |
| dansmith | clarkb: ack | 16:00 |
| clarkb | Most other pypi users will just install a 4 month old package (possibly with security vulnerabilities) without any warning | 16:00 |
| dansmith | but presumably this takes some time to clear then as well yeah? | 16:01 |
| clarkb | it has been hard to nail down. The last time we brought it up with them their resposne initially was "not our problem". Eventually we dug up enoguh logs to convince them there were issues and they discovered their backup backed had run out of disk and was even more out of date than anticipated | 16:02 |
| sean-k-mooney | this is py38 | 16:02 |
| sean-k-mooney | so the >=3.7 requirement shoudl not be a problem | 16:02 |
| sean-k-mooney | jinja2 also requires >=3.7 | 16:02 |
| clarkb | I personally think they shouldn't fallback to a backend that can serve pckages with known security bugs that have been fixed. They should just error for everyone | 16:02 |
| sean-k-mooney | locally i can install the tempest vnev target on 3.8 | 16:03 |
| clarkb | sean-k-mooney: yup agreed. Just calling out the other common ways this manifests and that those should be double checked | 16:03 |
| clarkb | sean-k-mooney: ya its a CDN issue so locality matters (and even then only one of a subset of your local cdn endpoints may have cached the bad data) | 16:03 |
| clarkb | But if they error for everyone their error rate would go up significantly | 16:03 |
| dansmith | clarkb: yeah this is ringing a bell now, I guess the error we see looks so "impossible requirement" that I assume it's our end | 16:04 |
| dansmith | but makes sense | 16:04 |
| dansmith | anyway, given the deadline this week I just wanted to raise the flag early | 16:05 |
| dansmith | hopefully it will clear before we get a big logjam | 16:05 |
| sean-k-mooney | ack so we just need to wait for the cdns to flush the cached data | 16:05 |
| clarkb | One thing that you can try is doing a PURGE http request against the resource that we think is cached wtih stale data | 16:05 |
| clarkb | I'm not sure if they still honor the PURGE requests but in the past the idea was if they got a PURGE request the CDN would rerequest the data | 16:06 |
| clarkb | `curl -X PURGE https://pypi.org/simple/markupsafe'` I think | 16:06 |
| clarkb | I'm super happy for someone else to interface with them though. My luck has always been poor. | 16:07 |
| clarkb | Most recently they told us that our caching proxies aren't going to work once they start signing indexes | 16:08 |
| sean-k-mooney | wont OPTIONS tell you what you can send it looks like PURGE is non standard | 16:08 |
| clarkb | we'll remove the caching proxy at that point and jobs will talk directly to pypi | 16:08 |
| clarkb | sean-k-mooney: ya its some cdn thing | 16:09 |
| sean-k-mooney | cache-control: max-age=600, public | 16:13 |
| sean-k-mooney | so it should expire after 10 minutes anyeay | 16:13 |
| dansmith | hope that's accurate, | 16:13 |
| dansmith | but I think that's telling the *client* to cache for ten minutes, not that the CDN will right? | 16:13 |
| sean-k-mooney | actully yes we woudl have to hit the backign webserver directly | 16:15 |
| sean-k-mooney | although with some cdns you can also configure it seperatly | 16:15 |
| sean-k-mooney | i guess we will see shortly | 16:15 |
| opendevreview | Emma Foley proposed openstack/devstack stable/xena: Respect constraints on tempest venv consistently https://review.opendev.org/c/openstack/devstack/+/855146 | 16:21 |
| *** jpena is now known as jpena|off | 16:42 | |
| *** efoley_ is now known as efoley | 17:24 | |
| *** pojadhav|ruck is now known as pojadhav|out | 17:53 | |
| opendevreview | Merged openstack/tempest master: Tempest test for checking unshelve to host https://review.opendev.org/c/openstack/tempest/+/841088 | 18:25 |
| opendevreview | Merged openstack/devstack master: Respect constraints on tempest venv consistently https://review.opendev.org/c/openstack/devstack/+/848482 | 22:53 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!