[09:33:49] (03CR) 10Ilias Sarantopoulos: [C: 03+1] "LGTM!" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/865713 (https://phabricator.wikimedia.org/T323023) (owner: 10AikoChou) [10:20:47] o/ question: I want to add a new pipeline for a new docker image in CI. I added it in the `.pipeline/config.yaml` but it won't run when i push a new commit. Does it need to be merged in main first or is there something else I should do? [10:24:41] isaranto: you also need to add the new pipeline to the integration/config repo, like https://gerrit.wikimedia.org/r/c/integration/config/+/849480 [10:28:34] aha, that was the missing piece, thanks Aiko! [10:41:27] (03CR) 10Ilias Sarantopoulos: [C: 03+1] "Nice addition!" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/866360 (https://phabricator.wikimedia.org/T323613) (owner: 10AikoChou) [10:54:17] Added the new pipeline here -> https://gerrit.wikimedia.org/r/c/integration/config/+/866560 [11:46:16] for some reason I messed it up with gerrit as commit msg was too long :shrug . so there is a new patch https://gerrit.wikimedia.org/r/c/integration/config/+/866570 [12:03:32] (03PS4) 10AikoChou: revertrisk: fix mwapi session host headers [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/865713 (https://phabricator.wikimedia.org/T323023) [12:04:42] (03CR) 10AikoChou: "Thanks for the review :)" [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/865713 (https://phabricator.wikimedia.org/T323023) (owner: 10AikoChou) [12:14:55] isaranto: o/ when you have a moment could you test the revscoring lua code for wrk? just to make sure it works :) [12:15:32] and then I'll merge the patch :) [12:16:52] (03CR) 10AikoChou: [C: 03+2] revertrisk: fix mwapi session host headers [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/865713 (https://phabricator.wikimedia.org/T323023) (owner: 10AikoChou) [12:21:02] (03Merged) 10jenkins-bot: revertrisk: fix mwapi session host headers [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/865713 (https://phabricator.wikimedia.org/T323023) (owner: 10AikoChou) [12:39:00] I'm doing it now [12:44:57] (03PS2) 10AikoChou: revertrisk: update to multilingual revertrisk model [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/861434 (https://phabricator.wikimedia.org/T323613) [12:49:36] aiko: I tested it on deployment server, works like a charm! [12:49:40] * isaranto afk lunch [12:54:44] (03Abandoned) 10Ladsgroup: Disable precaching for all wikis except enwiki [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/672881 (owner: 10Ladsgroup) [14:00:10] isaranto: nice! :) [14:07:17] klausman: o/ https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/866569 [14:10:23] looking [14:11:08] Do you want to go through with merge+deploy, or wait 'til Monday? [14:13:42] (03PS1) 10Ilias Sarantopoulos: blubber: create universal revscoring image [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/866591 (https://phabricator.wikimedia.org/T323586) [14:15:40] (03Abandoned) 10Ilias Sarantopoulos: blubber: create universal revscoring image [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/866591 (https://phabricator.wikimedia.org/T323586) (owner: 10Ilias Sarantopoulos) [14:16:23] Eh, it's experimental. +2'ing :) [14:17:14] Waiting for Jerkins to do its thing [14:19:55] 10Machine-Learning-Team, 10Patch-For-Review: Reduce number of published docker images for revscoring models - https://phabricator.wikimedia.org/T323586 (10isarantopoulos) At the moment I have created one image for all revscoring models and managed to run inference through that. We build an image of approx 1.5G... [14:22:25] aiko: Ok, docker image update is merged [14:22:59] klausman: thanks :) I'm gonna deploy it to staging now. [14:23:05] Ack. [14:24:50] Morning all! [14:25:02] \o [14:26:02] Looks like the structured data team have a model made by Diego they want us to deploy. We’ll talk about it next week at the team meeting. Xgboost so nothing too wild [14:29:48] 10Machine-Learning-Team, 10Patch-For-Review: Reduce number of published docker images for revscoring models - https://phabricator.wikimedia.org/T323586 (10isarantopoulos) My suggestion to proceed would be the following: - introduce new image, deploy and test it wherever we want - deprecate old files and pipeli... [14:38:30] klausman: sorry I didn't notice there was a typo :( https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/866600 [14:39:10] +2'd. And no worries :) [14:59:24] * klausman very late lunch [15:15:54] morning Chris /o [15:16:55] morning! [15:16:57] I still do it the other way around /o instead of o/ [15:17:38] I'm just a chatGPT now. Feel free to ask for management support as a poem or in Mandarin [15:18:42] 😂 [15:18:56] :D [15:19:47] isaranto: I mean, /o is valid. Think "hand on your forehead" [15:20:08] aAnd of course the two classics of \o/ and /o\ [15:20:29] being "yay!" and "oh noes!" [16:58:05] (03CR) 10Ilias Sarantopoulos: blubber: create universal revscoring image (031 comment) [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/865670 (https://phabricator.wikimedia.org/T323586) (owner: 10Ilias Sarantopoulos) [17:02:56] heya, I opened the patch for review. I have some extra thoughts I'll put on phabricator but in general I have the following two suggestions/thoughs we could discuss: [17:02:56] 1. use a pre-commit hook at least for linting/styling for consistency and then use that in CI [17:02:56] 2. start structuring code using some sw design patters: at the moment everything is great but and when things scale we will have to make multiple changes and fill our code with if/else statements. This is ofc a design design decision so I don't expect to get a specific response at the moment [17:06:09] I'm leaving u on this friday afternoon with my random thoughts, cya. on monday folks! [17:29:53] 10Lift-Wing, 10Machine-Learning-Team, 10Patch-For-Review: Test batch prediction for revert-risk model - https://phabricator.wikimedia.org/T323023 (10achou) The issue with missing responses in KServe was because we shared a client session between requests, the host header couldn't get updated correctly. We... [17:44:17] Sounds good, lets chat about them next week [18:17:19] (03PS3) 10AikoChou: Create a test folder and add lua scripts for wrk [machinelearning/liftwing/inference-services] - 10https://gerrit.wikimedia.org/r/866360 (https://phabricator.wikimedia.org/T323613) [19:45:57] have a good weekend folks! :)