| opendevreview | Joan Gilabert proposed openstack/hacking master: Add H302 check: import modules, not objects https://review.opendev.org/c/openstack/hacking/+/1011889 | 07:19 |
|---|---|---|
| *** elodille1 is now known as elodilles | 12:06 | |
| dansmith | trivial devstack change to stop forcing nova into an instance name template from 2010: https://review.opendev.org/c/openstack/devstack/+/1012535 | 13:54 |
| dansmith | we can change these on the fly now and want to migrate our default to something sane for 2012 (yes that's right) but devstack is forcing us for some ancient reason | 13:55 |
| frickler | wow, I never looked that deep down into devstack history, but that's pretty impressive. though not enlightening regarding the rationale https://review.opendev.org/c/openstack/devstack/+/2438/1/stack.sh#1046 | 14:01 |
| dansmith | yeah no idea why we've kicked that along for so long | 14:02 |
| opendevreview | Merged openstack/devstack-plugin-ceph master: Set RBD EC data pool defaults for services https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/1011533 | 14:02 |
| dansmith | maybe the actual original default in nova was even stupider? | 14:02 |
| frickler | hmm, do I dare doing the same git-blame deep-dive into the nova depths of time now? ;) | 14:03 |
| dansmith | IIRC nova used to use the actual DB row ID as the external ID of the instance | 14:52 |
| dansmith | I'm not sure what else the libvirt name might have been before it, except that maybe there was no default when xen was the pseudo default virt driver? | 14:53 |
| opendevreview | Merged openstack/devstack master: Stop overriding instance_name_template in config https://review.opendev.org/c/openstack/devstack/+/1012535 | 15:54 |
| opendevreview | Friedrich Zahn proposed openstack/tempest master: Fix endpoint filtering for missing region https://review.opendev.org/c/openstack/tempest/+/1012505 | 17:14 |
| gmaan | meeting time | 18:00 |
| gmaan | #startmeeting qa | 18:00 |
| opendevmeet | Meeting started Wed Oct 28 18:00:43 2026 UTC and is due to finish in 60 minutes. The chair is gmaan. Information about MeetBot at http://wiki.debian.org/MeetBot. | 18:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 18:00 |
| opendevmeet | The meeting name has been set to 'qa' | 18:00 |
| gmaan | we do not have wiki now for the agenda so I will go through it from meetings logs and later will prepare it on etherpad | 18:01 |
| gmaan | who all here today | 18:01 |
| gmaan | seems no one then I will skip the today meeting with action item on me to migrate the QA meeting etc to etherpad | 18:03 |
| gmaan | #action gmaan to migrate the a meeting or any other wiki to etherpad | 18:03 |
| gmaan | #endmeeting | 18:03 |
| opendevmeet | Meeting ended Wed Oct 28 18:03:32 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 18:03 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/qa/2026/qa.2026-10-28-18.00.html | 18:03 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/qa/2026/qa.2026-10-28-18.00.txt | 18:03 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/qa/2026/qa.2026-10-28-18.00.log.html | 18:03 |
| frickler | dansmith: I just found https://review.opendev.org/c/openstack/devstack/+/891213/1 and I think that that may still be interesting, maybe you can rebase? | 18:33 |
| frickler | gmaan: seems some of the centos-type jobs are broken, not sure if anyone is still around to try and fix them, otherwise we should reduce coverage? | 18:34 |
| gmaan | yeah, c-9 passing but c-10 failing https://zuul.openstack.org/builds?job_name=tempest-full-centos-9-stream&job_name=devstack-platform-centos-9-stream&job_name=devstack-platform-centos-10-stream&skip=0 | 18:35 |
| gmaan | I am not aware if anyone looking into it | 18:35 |
| gmaan | will send it on ML if no-one fix it during this week | 18:36 |
| cardoe | gmaan: frickler: Can we land https://review.opendev.org/c/openstack/devstack/+/983361 that'll improve our testing. We've rolled that setting out for OpenStack Helm as well now. | 18:44 |
| dansmith | frickler: oh man, I don't even have that git tree anymore, but I guess I can pull it down and try | 18:52 |
| dansmith | melwitt: I don't know why I never even followed up with your comments ^ | 18:55 |
| dansmith | I feel like I must have tested that locally but... | 18:55 |
| melwitt | heheh | 18:55 |
| dansmith | melwitt: does that bring up any memories? like should I just change it to your suggestion? | 18:58 |
| melwitt | dansmith: honestly don't really remember now but I'm guessing low-slow does not exist and log-slow does | 19:00 |
| dansmith | ohhh I didn't even spot that change.. you also specified a group though which I thought was the suggestion | 19:01 |
| dansmith | but yeah I see | 19:01 |
| dansmith | so almost certainly didn't test locally :P | 19:01 |
| melwitt | oh, yeah. the group wasn't really the main part. maybe it should not have a group but I assumed it would bc the others do | 19:02 |
| opendevreview | Dan Smith proposed openstack/devstack master: Log slow uwsgi requests https://review.opendev.org/c/openstack/devstack/+/891213 | 19:04 |
| opendevreview | Dan Smith proposed openstack/devstack master: Log slow uwsgi requests https://review.opendev.org/c/openstack/devstack/+/891213 | 19:05 |
| dansmith | frickler: ^ as tested as the original apparently ;P | 19:05 |
| opendevreview | Merged openstack/devstack stable/2026.2: Cap stable/2026.2 network, swift, volume api_extensions for tempest https://review.opendev.org/c/openstack/devstack/+/1006883 | 19:06 |
| opendevreview | Merged openstack/devstack stable/2026.2: Install glance_store S3 extra https://review.opendev.org/c/openstack/devstack/+/1012408 | 19:10 |
| opendevreview | Merged openstack/devstack master: Remove remaining ironic-lib https://review.opendev.org/c/openstack/devstack/+/1012199 | 19:10 |
| frickler | cardoe: I commented with a question just earlier | 19:45 |
| cardoe | frickler: I appreciate it. | 19:46 |
| cardoe | I'll comment on there as well. | 19:46 |
| cardoe | But you're not wrong that it's less than ideal but the blog post is wrong. | 19:47 |
| cardoe | So today when we have 4 uWSGI processes running for example, if 1 works and 3 fail... the other 3 spam no actively loaded app in the logs and we're only processing 1 request at a time. Zuul tests don't typically make that clear though because the service is somewhat responsive and you usually just take longer or fail to run time. | 19:48 |
| cardoe | With that change the child process instead exits when it fails to load the app and uWSGI just restarts it | 19:49 |
| cardoe | So that can lead to a lot of CPU churn if its stuck failing. | 19:50 |
| cardoe | Best case would be to get rid of lazy-app entirely | 19:50 |
| cardoe | But there's still a lot of libraries we've got that just aren't fork safe and load something in the global scope | 19:50 |
| cardoe | for example ovsdbapp that neutron uses | 19:50 |
| cardoe | Just importing that library and then forking your process causes problems | 19:51 |
| cardoe | glance or swift have one as well. | 19:51 |
| cardoe | So I've been proposing each OpenStack service drop lazy-app one at a time until I can catch them all. | 19:52 |
| cardoe | What I really need to do is make a community goal and track it. | 19:52 |
| opendevreview | Dan Smith proposed openstack/devstack master: Log slow uwsgi requests https://review.opendev.org/c/openstack/devstack/+/891213 | 20:49 |
| opendevreview | melanie witt proposed openstack/devstack master: Fix GMR collection task killing itself via pkill -f https://review.opendev.org/c/openstack/devstack/+/1012780 | 20:56 |
| opendevreview | melanie witt proposed openstack/devstack master: Fix GMR collection task killing itself and running too late https://review.opendev.org/c/openstack/devstack/+/1012780 | 23:36 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!