| opendevreview | Gregory Thiemonge proposed openstack/octavia stable/2025.1: Update diskimage-create.sh to use noble as default https://review.opendev.org/c/openstack/octavia/+/970341 | 08:31 |
|---|---|---|
| opendevreview | Ivan Anfimov proposed openstack/octavia-dashboard stable/2025.2: Optimize empty loadbalancer page https://review.opendev.org/c/openstack/octavia-dashboard/+/970375 | 13:14 |
| opendevreview | Ivan Anfimov proposed openstack/octavia-dashboard stable/2025.1: Optimize empty loadbalancer page https://review.opendev.org/c/openstack/octavia-dashboard/+/970377 | 13:14 |
| opendevreview | Merged openstack/octavia-dashboard master: Optimize empty loadbalancer page https://review.opendev.org/c/openstack/octavia-dashboard/+/969295 | 13:15 |
| opendevreview | Théo Van Gyzel proposed openstack/octavia master: API data mapping performance enhancements https://review.opendev.org/c/openstack/octavia/+/970404 | 14:59 |
| theovg | Hey all | 15:07 |
| theovg | In an effort to improve API performance, I've started a patch and I would like some input before proceeding with more work and cleanup | 15:07 |
| theovg | Essentially DB model -> API model instead of DB model -> Data model -> API model | 15:07 |
| theovg | https://review.opendev.org/c/openstack/octavia/+/970404 | 15:07 |
| theovg | I'm getting the following improvements with the three calls I've worked on for now (same DevStack, same data): | 15:07 |
| theovg | - openstack loadbalancer show : 3.93s to 0.08s | 15:07 |
| theovg | - openstack loadbalancer listener show : 4.22s to 0.05s | 15:07 |
| theovg | - openstack loadbalancer status show : 5.25s to 2.88s | 15:07 |
| opendevreview | Merged openstack/octavia-tempest-plugin master: Remove outdated comments from requirements files https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/962946 | 15:07 |
| opendevreview | Merged openstack/octavia-tempest-plugin master: Remove useless cover target https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/962947 | 15:07 |
| johnsom | #startmeeting Octavia | 16:01 |
| opendevmeet | Meeting started Wed Dec 10 16:01:00 2025 UTC and is due to finish in 60 minutes. The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot. | 16:01 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 16:01 |
| opendevmeet | The meeting name has been set to 'octavia' | 16:01 |
| gthiemonge | o/ | 16:01 |
| johnsom | Hello everyone | 16:01 |
| rcruise-redhat | Good morning! | 16:01 |
| johnsom | #topic Announcements | 16:02 |
| johnsom | The only announcement I have is we are getting into the holiday season in the US and I will not be available until the 7th. | 16:02 |
| johnsom | We can either cancel the weekly meeting for those weeks or someone else can host. | 16:03 |
| johnsom | What are your thoughts? | 16:03 |
| gthiemonge | I'll be on PTO too until January (except next week), so maybe we can cancel | 16:03 |
| johnsom | Ok, sounds good. I will send out an email to the discuss list later today. | 16:05 |
| johnsom | #topic Brief progress reports / bugs needing review | 16:05 |
| gthiemonge | ack | 16:05 |
| johnsom | I have been mostly consumed with downstream work and have not made much progress here this week. | 16:05 |
| gthiemonge | I found out that the 2025.1 jobs don't use Noble amphora image (but the control plan runs on Noble), so I backported: | 16:06 |
| gthiemonge | https://review.opendev.org/c/openstack/octavia/+/970341 | 16:06 |
| johnsom | Ack, thanks. | 16:07 |
| gthiemonge | (I was looking at backporting a fix for haproxy 2.8, and the CI job didn't seem impacted by the bug, because it's still using an older version) | 16:07 |
| gthiemonge | https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/956752 | 16:08 |
| gthiemonge | ^ is not supposed to pass without a backport | 16:09 |
| johnsom | Does that need a depends-on? | 16:09 |
| gthiemonge | i'm w-1 right now, it may need a depends on a backport that doesn't exist yet :D | 16:10 |
| johnsom | Ok, if there are no other updates, we can move on to open discussion | 16:12 |
| johnsom | #topic Open Discussion | 16:13 |
| gthiemonge | earlier today, someone mentioned a proposal to improve API performance | 16:13 |
| gthiemonge | https://review.opendev.org/c/openstack/octavia/+/970404 | 16:14 |
| gthiemonge | results look good, but I'm not sure we want db queries directly in the API code (or maybe it's fine) | 16:14 |
| johnsom | I am guessing that his related to the proposal to flatten the data model? | 16:14 |
| gthiemonge | I don't think it is | 16:15 |
| johnsom | Oh, is it related to the performance drop in sqlalchemy? | 16:15 |
| johnsom | versions > 1.4 had a pretty big drop | 16:16 |
| gthiemonge | it skips some internal data conversions | 16:16 |
| theovg | Hi, I'm the owner of the proposal; DB queries I was thinking of offloading. I'm looking for opinions on the general idea before putting more work into it | 16:16 |
| theovg | It is mostly about bypassing to_data_model and from_data_model | 16:16 |
| johnsom | Ah, yeah, so probably related to the "flatten data model" proposal. (I have not yet looked at the patch) | 16:17 |
| johnsom | Currently the data model pulls in everything, recursively, and then picks what it needs from that data. It's horribly inefficient but simple. We have discussed that doing this does not have a lot of value and we should make all of the objects flat and not recurse as it's not really needed or could be done in the few cases it is needed. | 16:19 |
| johnsom | A quick look at the patch, I would encourage us to keep all of the sql specific code in the octavia/db/repositories.py file. | 16:21 |
| johnsom | This has helped us in the past when sqlalchemy makes breaking changes for example. | 16:21 |
| theovg | That was my observation; the patch I proposed lays out a parallel processing path "from_db_obj" which bypasses the data model system completely and showing great results on my benchmarks all without breaking any tests | 16:22 |
| gthiemonge | yeah in the proposal, skipping the conversions to the data model removes the issue with the recursion of the objects | 16:22 |
| theovg | Definitely agree on the query code | 16:22 |
| johnsom | I will see if I can give you initial comments this week to not block your progress. I think you are on the same path we agreed needed to happen a while ago. | 16:23 |
| theovg | Thank you! | 16:23 |
| johnsom | Ok, we have a path forward here. Any other topics today? | 16:25 |
| gthiemonge | nothing from me | 16:25 |
| rcruise-redhat | none from me | 16:26 |
| johnsom | All right, thank you everyone! | 16:26 |
| johnsom | #endmeeting | 16:26 |
| opendevmeet | Meeting ended Wed Dec 10 16:26:26 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 16:26 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/octavia/2025/octavia.2025-12-10-16.01.html | 16:26 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/octavia/2025/octavia.2025-12-10-16.01.txt | 16:26 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/octavia/2025/octavia.2025-12-10-16.01.log.html | 16:26 |
| theovg | Thanks again, have a good one | 16:26 |
| rcruise-redhat | Thanks all, enjoy the rest of your week! | 16:27 |
| opendevreview | Richard Cruise proposed openstack/octavia master: Fix spurious logging message in amphora-agent API https://review.opendev.org/c/openstack/octavia/+/970421 | 16:28 |
| opendevreview | Richard Cruise proposed openstack/octavia master: Fix spurious logging message in amphora-agent API https://review.opendev.org/c/openstack/octavia/+/970421 | 16:28 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!