[00:25:36] Daimona: OK if I alter https://gerrit.wikimedia.org/r/c/operations/mediawiki-config/+/647091/ to only set WRITE_OLD for now, and set WRITE_BOTH for Beta Cluster? [00:26:13] As a test? Sure [00:26:22] (03PS21) 10DannyS712: FunctionCommentSniff: Move some sniffs to new separate sniff [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/595737 (https://phabricator.wikimedia.org/T250958) [00:27:54] Daimona: I'll wait for the backport window to be over. [00:28:33] Sure, no worries, and thank you. Of note, this is also not going to have any effect in prod until next week [00:28:59] (03PS22) 10DannyS712: FunctionCommentSniff: Move some sniffs to new separate sniff [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/595737 (https://phabricator.wikimedia.org/T250958) [00:29:18] No, indeed, but running the script in Beta Cluster today and feeling confident that we can run it later would be good. [00:29:24] Given the impending no-deploy time. [00:31:07] (03PS23) 10DannyS712: FunctionCommentSniff: Move some sniffs to new separate sniff [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/595737 (https://phabricator.wikimedia.org/T250958) [00:33:27] (03CR) 10DannyS712: "This change is ready for review." [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647072 (https://phabricator.wikimedia.org/T252914) (owner: 10DannyS712) [00:39:18] 10MediaWiki-Codesniffer, 10Patch-For-Review, 10User-DannyS712: Add sniff to disallow specific tags with no text after them - https://phabricator.wikimedia.org/T252914 (10DannyS712) a:03DannyS712 [00:41:35] James_F: Oh, right. Perhaps it can be backported, then? So we can run it this week, and there's more time to fix errors etc.? [00:43:10] Daimona: Nah, it's fine, we can wait a bit. [00:43:18] Daimona: We'll know today if it works. [00:43:22] Sure [00:43:57] legoktm: CI builds its images from the Wikimedia apt repo with the version of wikidiff2 there in it. Not sure how we can move CI forwards neatly without changing the version new appservers would also get? [00:45:49] the new version only matters for PHP 8.0 and we get those packages from sury, not our apt repo [00:45:57] and sury doesn't even have wikidiff2 [00:45:58] Oh, point. [00:46:12] But where do we get wikidiff2 from for that? [00:46:30] Or do we just silently not have that module on those images and I've never noticed? [00:46:30] it just isn't installed, any tests that might require it (I doubt there are any) just skip [00:46:34] yep [00:46:36] Ha. [00:46:43] Well, that makes that part of life easier. :-) [01:12:40] Daimona: Good news. Found a bug. [01:13:29] Lovely. [01:13:30] Just one? [01:13:42] Reedy: So far. PHP has the habit of exiting on errors. ;-) [01:13:47] for the record, fixing T268200 might help with finding more [01:13:48] T268200: Beta cluster logstash down - https://phabricator.wikimedia.org/T268200 [01:13:49] Reedy: That's the good news [01:13:58] * James_F grins. [01:15:24] "All supported databases have substring functions that behave the same for positive $startPosition and non-negative $length, but behaviors differ when given 0 or negative $startPosition or negative $length. The simplest solution to that is to just forbid those values." [01:15:35] Yay... [01:16:15] Gotta love MW. [01:16:26] Not sure why, but it's the law. You *have* to. [01:17:14] Welp, seems like we cannot get the starting segment of a string with our DBAL [01:18:50] I'm not really sure there's a solution, because we need that in the UPDATE clause. Unless we want to replace a single mass-update with $batchSize different updates, of course [01:19:06] Eurgh. [01:19:11] I guess we're going to have to? [01:19:22] No, let's fix the DBAL [01:19:29] Or we can patch MW to allow it temporarily in mysql and then ban it again? [01:19:42] OK, yeah, we can permanently fix it, but that's $EFFORT. ;-) [01:19:55] Yeah I'd have to quantify the $effort there [01:20:10] (Which is not quite easy when it's 2:20AM eheh) [01:20:19] No rush. [01:21:50] And actually, wait. That line was introduced when we were still supporting Oracle and MSSQL [01:22:11] So perhaps we can just remove it if the 3 remainings DBMSs work fine with zeros [01:22:35] If. [01:28:13] So, MySQL wants 1 as start position [01:28:33] Same for Postgres [01:29:13] And SQLite too [01:29:24] (03PS1) 10BryanDavis: builder: support cross variant copying for builder.requirements [blubber] - 10https://gerrit.wikimedia.org/r/647120 (https://phabricator.wikimedia.org/T263597) [01:30:26] Oh, OK, that's not too bad then. [01:30:40] How good is our test suite for DBAL stuff? [01:31:07] And so, after matlab and lua, here's another situation where things can be 1-based [01:31:18] Tsk. [01:31:21] :-) [01:31:27] I think the tests might as well be fine, it's just the error message that might be improved a bit [01:31:44] Like: "String indexes are 1-based" if the start pos is 0 [01:32:06] Well, anyway, https://gerrit.wikimedia.org/r/c/mediawiki/extensions/AbuseFilter/+/647121 [01:35:30] And https://gerrit.wikimedia.org/r/c/mediawiki/core/+/647122 is for the error [01:40:17] (03CR) 10BryanDavis: "marxarelli: this is working with my blubber.yaml and outputs the Dockerfile I hoped for. I need to write some more tests and update the op" [blubber] - 10https://gerrit.wikimedia.org/r/647120 (https://phabricator.wikimedia.org/T263597) (owner: 10BryanDavis) [01:42:57] Well, time to go. James_F: thank you for running the script, I'm curious to see how many bugs are yet to be found :-) [01:45:25] * James_F grins. [01:51:05] (03CR) 10BryanDavis: builder: support cross variant copying for builder.requirements (031 comment) [blubber] - 10https://gerrit.wikimedia.org/r/647120 (https://phabricator.wikimedia.org/T263597) (owner: 10BryanDavis) [02:34:19] (03PS3) 10Reedy: Bump composer/semver versions [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) [02:34:44] (03CR) 10Jforrester: [C: 03+2] Bump composer/semver versions [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) (owner: 10Reedy) [02:36:25] (03Merged) 10jenkins-bot: Bump composer/semver versions [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) (owner: 10Reedy) [02:52:33] (03CR) 10Reedy: Bump composer/semver versions (031 comment) [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) (owner: 10Reedy) [02:53:21] (03CR) 10Jforrester: Bump composer/semver versions (031 comment) [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) (owner: 10Reedy) [02:55:06] (03CR) 10Reedy: Bump composer/semver versions (031 comment) [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) (owner: 10Reedy) [03:01:36] (03CR) 10Jforrester: Bump composer/semver versions (031 comment) [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647076 (https://phabricator.wikimedia.org/T269068) (owner: 10Reedy) [03:07:35] 10Release-Engineering-Team-TODO (2020-10-01 to 2020-12-31 (Q2)), 10Platform Team Workboards (Green), 10Release, 10Train Deployments: Old revision warning box is added twice on page view if old rev served from cache - https://phabricator.wikimedia.org/T269727 (10Pchelolo) [03:08:18] 10Release-Engineering-Team-TODO (2020-10-01 to 2020-12-31 (Q2)), 10Platform Team Workboards (Green), 10Release, 10Train Deployments: Old revision warning box is added twice on page view if old rev served from cache - https://phabricator.wikimedia.org/T269727 (10Pchelolo) [03:24:41] James_F if you have a minute, would you mind reviewing https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GlobalWatchlist/+/647079/ ? [03:50:19] DannyS712: Done. [03:54:01] thanks so much - I'll test again once it merges [03:55:04] also, anything I can review for you? [05:41:53] (03PS2) 10BryanDavis: builder: support cross variant copying for builder.requirements [blubber] - 10https://gerrit.wikimedia.org/r/647120 (https://phabricator.wikimedia.org/T263597) [05:54:32] 10Release-Engineering-Team-TODO (2020-10-01 to 2020-12-31 (Q2)), 10MediaWiki-Parser, 10Patch-For-Review, 10Platform Team Workboards (Green), 10Regression: Old revision warning box is added twice on page view if old rev served from cache - https://phabricator.wikimedia.org/T269727 (10DannyS712) [07:44:41] (03CR) 10Thiemo Kreuz (WMDE): [C: 03+1] "Looks fine when I compare it with what I wrote in T252914#6142288. Also, nothing is found in the test run, which is good, I think." [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647072 (https://phabricator.wikimedia.org/T252914) (owner: 10DannyS712) [08:21:51] (03CR) 10Thiemo Kreuz (WMDE): [C: 03+1] "I think I find a small issue in the auto-fix. Looks really good otherwise." (033 comments) [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647070 (https://phabricator.wikimedia.org/T242762) (owner: 10DannyS712) [08:42:21] (03CR) 10Thiemo Kreuz (WMDE): "Looks good, but …" (031 comment) [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647071 (https://phabricator.wikimedia.org/T268509) (owner: 10DannyS712) [08:49:39] (03PS4) 10Hashar: jjb: Provide PHP 7.3 and 7.4 version of PHPUnit coverage publish job [integration/config] - 10https://gerrit.wikimedia.org/r/628589 (owner: 10Jforrester) [08:49:41] (03PS1) 10Hashar: jjb: use same image in coverage publish job [integration/config] - 10https://gerrit.wikimedia.org/r/647194 [08:51:24] (03CR) 10Hashar: "I have changed it to a job template. In order to rely on a single image, I have introduced an intermediate step that replaces releng/php72" (032 comments) [integration/config] - 10https://gerrit.wikimedia.org/r/628589 (owner: 10Jforrester) [08:52:07] (03CR) 10Hashar: "Makes it possible to use a single docker image variable when the job is converted to a job template with https://gerrit.wikimedia.org/r/c" [integration/config] - 10https://gerrit.wikimedia.org/r/647194 (owner: 10Hashar) [08:52:25] (03CR) 10Thiemo Kreuz (WMDE): [C: 03+1] "I love it! Thanks for the update. I think I will merge this within the next days when nobody objects." [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/639887 (https://phabricator.wikimedia.org/T265740) (owner: 10Umherirrender) [08:56:11] (03CR) 10Hashar: [C: 03+2] "Lets do it! The first use case will be for a new job that solely triggers on integration/quibble changes and would thus have the latest c" [integration/config] - 10https://gerrit.wikimedia.org/r/644485 (https://phabricator.wikimedia.org/T225218) (owner: 10Awight) [08:57:58] (03Merged) 10jenkins-bot: Image for quibble serving from apache [integration/config] - 10https://gerrit.wikimedia.org/r/644485 (https://phabricator.wikimedia.org/T225218) (owner: 10Awight) [09:16:03] (03CR) 10Kosta Harlan: [C: 03+1] "Changes in PS8-PS-10 look good to me, thanks Adam!" [integration/quibble] - 10https://gerrit.wikimedia.org/r/644178 (owner: 10Kosta Harlan) [09:30:37] (03PS1) 10Hashar: (DO NOT SUBMIT) [integration/config] - 10https://gerrit.wikimedia.org/r/647199 [09:33:22] (03CR) 10Awight: Reformat files with black (034 comments) [integration/quibble] - 10https://gerrit.wikimedia.org/r/644178 (owner: 10Kosta Harlan) [09:34:15] kostajh: awight: in case one of you is around. I have tried to build the quibble-apache image and it is missing a php.ini file for COPY ./php-fpm/php.ini /etc/php/7.2/fpm/php.ini [09:35:36] Oh [09:35:41] I can just grab it from quibble it seems [09:37:59] (03CR) 10Kosta Harlan: [C: 03+1] Reformat files with black (031 comment) [integration/quibble] - 10https://gerrit.wikimedia.org/r/644178 (owner: 10Kosta Harlan) [09:38:11] (03PS1) 10Hashar: dockerfiles: quibble-apache missed php.ini file [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) [09:39:13] (03CR) 10Hashar: [C: 03+1] "Copied from integration/quibble . It was not added in the previous chance since we had .gitignore igoring '*.ini'. The image builds now " [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [09:43:11] (03CR) 10Awight: "Nice find!" [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [09:44:08] awight: mind +2ing it? I will build the container [09:44:15] it managed to build locally at lesat! [09:47:18] (03PS1) 10Thiemo Kreuz (WMDE): Avoid slow strcasecmp() where not necessary [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647205 [09:48:36] (03PS11) 10Awight: Reformat files with black [integration/quibble] - 10https://gerrit.wikimedia.org/r/644178 (owner: 10Kosta Harlan) [09:48:38] (03PS1) 10Awight: Subprocess argument fixes [integration/quibble] - 10https://gerrit.wikimedia.org/r/647206 [09:49:07] hashar: Looks like I don't have CR+2 in integration/config. Probably for the better ;-) [09:49:18] (03CR) 10Awight: [C: 03+1] "Would merge if I could!" [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [09:51:31] (03CR) 10Awight: Reformat files with black (031 comment) [integration/quibble] - 10https://gerrit.wikimedia.org/r/644178 (owner: 10Kosta Harlan) [09:51:33] (03CR) 10Kosta Harlan: [C: 03+1] dockerfiles: quibble-apache missed php.ini file [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [10:03:58] (03CR) 10Ammarpad: [C: 03+1] Avoid slow strcasecmp() where not necessary [tools/codesniffer] - 10https://gerrit.wikimedia.org/r/647205 (owner: 10Thiemo Kreuz (WMDE)) [10:21:03] awight: :D [10:21:17] (03CR) 10Hashar: [C: 03+2] dockerfiles: quibble-apache missed php.ini file [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [10:22:37] (03Merged) 10jenkins-bot: dockerfiles: quibble-apache missed php.ini file [integration/config] - 10https://gerrit.wikimedia.org/r/647202 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [10:22:59] and it fails :/ [10:22:59] https://phabricator.wikimedia.org/P13536 [10:23:13] INFO:backend.DevWebServer:Starting external webserver [10:23:18] File "/usr/local/lib/python3.5/dist-packages/quibble/backend.py", line 393, in start [10:23:19] server_cmd, [10:23:19] UnboundLocalError: local variable 'server_cmd' referenced before assignment [10:23:19] hehe [10:23:34] cause well that is using quibble 0.0.45 [10:26:22] hashar: can you let me know a good time to merge https://gerrit.wikimedia.org/r/c/integration/config/+/645143 and the associated puppet change [10:28:17] !log Successfully tagged docker-registry.discovery.wmnet/releng/quibble-apache:0.0.1 T225218 [10:28:20] Logged the message at https://wikitech.wikimedia.org/wiki/Release_Engineering/SAL [10:28:20] T225218: Consider httpd for quibble instead of php built-in server - https://phabricator.wikimedia.org/T225218 [10:28:39] jbond42: oh yeah forgot about this one [10:29:03] jbond42: the image rebuild is merely to have the parallel_spec gem to be preinstalled in the image isn't it ? [10:29:14] yes exactly [10:29:25] so yeah lets do that ;] [10:29:33] ok cool ill merge the puppet one now [10:30:27] ok thats merged [10:31:35] (03CR) 10Hashar: "Deployed and a first run is https://integration.wikimedia.org/ci/job/integration-quibble-apache-fullrun/1/ There is certainly going to n" [integration/config] - 10https://gerrit.wikimedia.org/r/647199 (owner: 10Hashar) [10:31:41] jbond42: rebuilding ;) [10:31:49] (03CR) 10Hashar: [C: 03+2] dockerfiles/operations-puppet: add parallel_spec to Rakefile [integration/config] - 10https://gerrit.wikimedia.org/r/645143 (owner: 10Jbond) [10:31:55] hashar: great thanks <3 [10:32:34] jbond42: possibly we could even split the job and have the profile specs to be run in a standalone job [10:32:57] (03Merged) 10jenkins-bot: dockerfiles/operations-puppet: add parallel_spec to Rakefile [integration/config] - 10https://gerrit.wikimedia.org/r/645143 (owner: 10Jbond) [10:33:03] but that means parallelism of all the rake tasks would be configured in the CI config instead of via the Rakefile so it is probably counterintuitive for most folks [10:33:32] hashar: i think im going to see about having some type of goal for next Q to take a good look at CI for the puppet repo [10:33:52] i think there is alrready some stuff we can do in rake to help inprove [10:35:00] i.e. we currently do modeules.each {|modules| system("cd #{modules}; rake parallel_spec")` [10:35:23] changing the to use parallel_spec directly instead of spawning rake for every module will help [10:35:43] most probably yeah [10:36:00] also useing parallel_spec will also mean we can (in theory) add additional cores to help [10:46:54] (03PS2) 10Hashar: Quibble full run with Apache [integration/config] - 10https://gerrit.wikimedia.org/r/647199 (https://phabricator.wikimedia.org/T225218) [10:48:01] (03CR) 10Hashar: "I have made the job to just run api-testing, qunit and selenium stages. An example run is ongoing at: https://integration.wikimedia.org/c" [integration/config] - 10https://gerrit.wikimedia.org/r/647199 (https://phabricator.wikimedia.org/T225218) (owner: 10Hashar) [10:48:29] awight: kostajh: the quibble apache fullrun is running at https://integration.wikimedia.org/ci/job/integration-quibble-apache-fullrun/3/console :] Only running qunit/api-testing/selenium though [10:48:34] we will see how it behaves [10:50:12] (03PS1) 10Hashar: jjb: use operations-puppet 0.8.0 for parallel_spec [integration/config] - 10https://gerrit.wikimedia.org/r/647213 [10:50:19] jbond42: Successfully tagged docker-registry.discovery.wmnet/releng/operations-puppet:0.8.0 [10:50:36] hashar: thanks ill kick of a run now to test [10:50:45] jbond42: we need to update the jenkins job [10:51:42] hashar: oh yes i can do that now [10:51:54] (03CR) 10Hashar: [C: 03+2] "operations-puppet-tests-buster-docker updated :)" [integration/config] - 10https://gerrit.wikimedia.org/r/647213 (owner: 10Hashar) [10:51:59] jbond42: done :] [10:52:03] :) thanks [10:52:41] argh [10:53:03] jbond42: I haven't seen the time, I have to rush to school to get my kids. Should I roll it back or is that going to work ? [10:53:04] ;) [10:53:06] sorry [10:53:41] (03Merged) 10jenkins-bot: jjb: use operations-puppet 0.8.0 for parallel_spec [integration/config] - 10https://gerrit.wikimedia.org/r/647213 (owner: 10Hashar) [10:53:42] do you have 60secs for the job to complete :) [10:53:46] yes [10:53:47] :) [10:54:30] I should really migrate all that crap to use scap instead [10:54:46] awight: did you test qunit with the apache backend? I didn't, and I see the job that hashar crafted is failing [10:54:52] hashar: thanks for making the job :) [10:54:58] jbond42: looks good ;) [10:55:09] yes looks good to me thx [10:55:17] \o/ [10:56:15] kostajh: yeah 00:03:04.359 09 12 2020 10:50:10.151:WARN [middleware:karma]: Invalid file type (), defaulting to js. [10:56:17] :\ [10:56:30] or something else really [10:56:43] anyway kids time and lunch etc. Be back later [11:05:03] kostajh: I don't remember testing qunit specifically /o\ [11:06:28] The correct URL does get wired into the qunit config. I wonder... [11:10:12] Seems like chrome-headless is the first thing to crash, with a "script error"? [11:14:10] +1 I can reproduce the error locally. [11:17:06] https://integration.wikimedia.org/ci/job/integration-quibble-apache-fullrun/3/artifact/log/apache-access.log , doesn't this have to be "/w/index.php" in CI? [11:20:00] Well the regular job has the same line: 10:57:19 INFO:backend.PhpWebserver:[Wed Dec 9 09:57:19 2020] 127.0.0.1:48210 [200]: //index.php?title=Special:JavaScriptTest/qunit/export [11:38:19] awight: maybe we are ùissing some env variables [11:39:11] or something is off in the apache conf. Guess one can spawn quibble with -c bash and then look at Special:JavaScriptTest in a browser [11:39:17] maybe it dies with a fatal error [11:40:15] I'm pretty suspicious of the missing /wiki vs /w distinction, but the browser tests work... [11:41:34] hasharLunch: Nice tip, thanks! [11:42:21] the sscript error might just be chromium