[00:08:54] not really related, but some parts are redirecting https -> http and then back to https :S [00:21:18] oh, actually that does mean something. Because cookies given out over https can be read from http. But still not relevant to the current problem :P [00:21:36] err, cookies given out over https can *not* be read from http [08:20:14] ebernhardson: when running make cirrussetup i get the error service "mediawiki" is not running container #1" [09:43:53] ejoseph: connectivity issues? [09:45:34] it took time to restart [09:45:43] kindly send me the meet link [09:45:53] https://meet.google.com/ukb-kgxq-gvq [11:20:04] Lunch [13:13:41] errand [13:58:24] Greetings [14:33:36] o/ [14:58:10] \o [15:09:11] ejoseph: hmm, unexepected :S I'll have to reproduce i suppose [15:09:29] Ok thanks [15:18:28] this makefile is for a different version of docker than I have :S [15:19:13] hax'd it up, hopefully reproduces same errors [15:20:01] hmm, different but probably similar 'No container found for mediawiki_1' [15:30:23] jossorry finding my self entirely distracted...will figure this out today but probably not before office hours [15:30:29] ejoseph: ^ [15:41:06] this script also uses odd commands that don't exist on linux distributions, was this written to only run on macos? [15:42:03] i suppose i don't know what the cross-distribution way would be to open a link in a browser. debian uses x-www-browser but not sure about others [15:46:00] gehel or anyone else, looks like "Monthly Tech Dept office hours" and "Talk to the Search Platform Team" are at the same itme, which should I attend? [15:46:25] inflatador: your choice! [15:47:15] * inflatador removes lucky half dollar from desk drawer [15:47:46] ebernhardson: I think it was written for Mac [15:48:39] OK, tech department wins [15:49:11] ejoseph: if you run this does a browser open: python3 -c 'import webbrowser; webbrowser.open("https://en.wikipedia.org")' [15:49:28] or more accurately, does the normal and expected browser open and not some random thing :) [16:02:01] Search Team call is NOW [16:02:20] We're in office hours now, with lots of visitors!! [16:02:37] https://meet.google.com/vgj-bbeb-uyi [16:58:37] quick workout, back in ~30 [18:02:20] inflatador, ryankemper: SDAW meeting: https://meet.google.com/qho-jyqp-qos [18:02:40] OMW [18:58:08] grabbing lunch, probably won't make the unmtg [19:23:43] and back [19:53:46] lunch [19:58:25] sorry, been back [20:12:33] quick break, back in ~15 [20:26:21] back [20:42:22] bck [20:56:17] hmm, while i have the mediawiki-cirrus-docker running, i suspect the differences in docker between macos and linux are going to make things difficult. In particular i had to resolve issues around wrong file permissions that likely depend on how volume mounts work [22:17:49] i'm not really figuring out how the permissions are supposed to work here :S Everything is written from the user account, but then the web server runs under `nobody`. How is nobody supposed to be able to write the sqlite file created by the real user account :S [22:21:23] ebernhardson: I can confirm that Docker Desktop on MacOS does completely different things with bind mount permissions than any other Docker-like runtime. [22:21:51] good things, but things that will bite you when you move from macos to anywhere else [22:22:57] bd808: yea my reading so far suggests things would be easier if linux did the way osx does, but the end result is "works on my machine" which docker was supposed to address :P [22:23:58] My tour of this looks something like: T295318, T296046, T295823 [22:23:58] T296046: Allow build time control of effective UID/GID for runtime in Blubber generated Dockerfile - https://phabricator.wikimedia.org/T296046 [22:23:59] T295318: Fix Toolhub developer environment tooling to work on Linux hosts - https://phabricator.wikimedia.org/T295318 [22:23:59] T295823: Failures updating python packages in dev environment following Linux Docker improvements - https://phabricator.wikimedia.org/T295823 [22:24:43] bd808: thanks! seems plausible some good referneces in here [22:30:06] When CI gets fixed so we can land that blubber change that was reverted I think there will be a complete solution for dev env use.