[00:33:15] @posix_memalign, @cosmicalpha, and @thewwrnerdguy: so I've been able to determine that if we have Babel userboxes on our global user page, on local wikis without the Babel extension enabled, the userboxes are disorderly. Where Babel is enabled on those wikis, it looks proper. I'm wondering if we could enable Babel as a default extension (can be disabled by local bureaucrats). Thoughts, and should I open a Phorge task to discuss this? [00:37:05] does babel do anything that Lua can't do? [00:38:28] I would just suggest we replace the userboxes away from bable instead to not rely on that, if possible. [00:40:07] loginwiki doesn't even have babel [00:42:01] [01:13:55] I think it's enabled on Loginwiki. Let me check [01:14:33] but if we can resolve the issue with a Lua module on Loginwiki, definitely open to that [01:15:24] fwiw https://login.miraheze.org/wiki/Special:ManageWiki/extensions#mw-section-parserhooks [06:03:57] ...which is exactly what wwr linked [10:11:09] I think we could actually remove babel from loginwiki tbh [10:40:07] [1/2] Ok I think I figured out one cause of the token problem. https://issue-tracker.miraheze.org/T15258#317009 [10:40:07] [2/2] Now the question is whether we are okay with me deploying a patch for that line in prod to see if things get better. [10:48:17] @posix_memalign we've had custom domains work on beta before [10:48:20] I think you're overthinking it [10:48:57] Ok I'll try to add the custom domain manually then [10:50:06] _remembers the tagline of last nights theatre: the closer you look, the less you see_ [10:52:09] Cloudflare expects a fallback origin for custom host names. Not sure what that would be. If it worked in the past using this route I assume there would already be some setup. [10:53:08] @posix_memalign still overthinking it [10:53:23] You can just add it to the prod saas setup [10:53:33] There should be an origin rule already there [10:57:13] I don't see the origin rules [10:57:53] But we did have it working before [10:58:13] I'm guessing you're saying I could do a URL rewrite so that requests that go to the custom domain are automatically rewritten to cd.mirabeta.org? [10:58:55] No [10:58:56] That is not what I said [10:59:04] There are a separate category of rules you can do to point them to a specific domain and use test151 [10:59:23] I think that will work [11:08:04] I made a origin rule that rewrites the SNI from test.strinova.org to cd.mirabeta.org. Domain misconfigured is as far as I got. [11:09:02] Hmmm. The js might have been broken due to a bad change. Lemme try again. [11:11:20] Still showing domain misconfigured. I guess I might be missing the use test151 step somewhere. [11:25:15] Did you add it in ManageWiki @posix_memalign ? [11:25:26] It got as far as mediawiki if you managed that [11:30:00] [1/2] Yeah. CSS is all broken bu the value did get set [11:30:01] [2/2] https://cdn.discordapp.com/attachments/1006789349498699827/1539959683937468516/image.png?ex=6a8836b7&is=6a86e537&hm=d9d04c1b3e19cd08adef4ed669af8b0922324c3a70bb0dd41ede760c7fce1ea6& [11:40:04] Hmm [11:40:11] Can't really debug as mobile from a train [11:41:10] We're not in a rush anyway. It's been an issue for months and I only made headway recently. [11:51:50] @posix_memalign I won't be able to look personally for like a week [11:51:59] Also https://github.com/miraheze/mw-config/pull/6502 is simple if you want an easy PR [11:52:45] I had an outstanding CSP PR which can probably be deployed together with it https://github.com/miraheze/mw-config/pull/6496 [11:54:17] Go for it [11:54:22] I will have no shell for a week [11:54:30] I approved your PR though [11:54:36] You can merge and deploy both [12:00:06] Seems to be working on beta. It's gonna be hard to test per-wiki overrides so I'm just going to deploy and see what happens. [12:02:53] Per-wiki overrides working, and we no longer have the domain of a random NSFW game in our CSP. [12:13:32] Though I think I should keep pushing for certain longstanding usability issues like CentralAuth cookie, post-install maintenance script failures, lacking a loadout system, etc. Just like how Random is still trying to get the server for speedscope. [12:19:47] Yes [12:19:50] I agree [13:10:06] Wait what [13:10:11] Which one was that [13:10:56] fallback origin is just an origin [13:11:09] so basically any dns record that points to the cf load balancer [13:33:50] can someone shoot at deploying https://github.com/miraheze/phorge-extensions/pull/73 to phorge for a bit and seeing if this time it works (likely not but hey) [13:35:46] If no one does before next Tuesday then I will @pixldev [13:35:57] I hope you're not waiting that long [13:36:02] What happened? [13:36:04] sure hope it wont take til tuesday but appreciate it? [13:36:13] I'm on leave [13:36:15] with the PR above or in general? [13:36:19] oh with rhinos [13:36:22] The wiki went down. [13:36:26] oh [13:36:29] Did it [13:36:33] that's not jolly [13:36:38] You should be more clear with what happened [13:36:49] Nothing is alerting [13:36:58] So what wiki if you're still having issues [13:37:07] It went away. [13:37:11] icinga isn't even screaming yeah. Is it a 50X or another error [13:37:20] 503, cloudflare. [13:37:59] It screamed [13:38:09] @paladox you around? [13:38:21] Or @abaddriverlol [13:39:12] yes? [13:39:33] @paladox db161 is showing signs of grumpiness [13:39:35] Per icinga [13:39:39] It's up [13:39:48] But probably a good time to have a look at what's running [13:39:50] yeh load went down [13:40:11] Is @orduin's script still there @paladox [13:40:25] That was checking what was running when critical load [13:40:33] We should make that something proper [13:41:37] doesn't look like it was running [14:23:42] I will today [14:24:32] Won't be until later though as I am currently out. [14:25:07] no worries [14:25:33] im OOO from about an hour from now to 6 or so hours later [15:28:29] @pixldev merged [15:28:32] and deployed [15:30:51] doesn't seem to work [15:46:58] seems to be defined again in https://github.com/phorgeit/phorge/blob/master/src/applications/auth/provider/PhabricatorAuthProvider.php#L517 [15:47:17] also puppet failed on phorge171 [19:32:42] Check `/var/log/mysql/proclist` it runs as a systemd script every minute (`mariadb-generate-processlist.service` and `.timer`) [19:34:43] oh [19:34:45] [1/7] > -rw-r--r-- 1 root mysql 6.3K Aug 19 11:20 proclist_20260819_112015.txt [19:34:46] [2/7] > -rw-r--r-- 1 root mysql 32K Aug 19 11:21 proclist_20260819_112118.txt [19:34:46] [3/7] > -rw-r--r-- 1 root mysql 117 Aug 19 11:22 proclist_20260819_112223.txt [19:34:46] [4/7] > -rw-r--r-- 1 root mysql 72K Aug 20 13:34 proclist_20260820_133450.txt [19:34:47] [5/7] > -rw-r--r-- 1 root mysql 89K Aug 20 13:35 proclist_20260820_133550.txt [19:34:47] [6/7] > -rw-r--r-- 1 root mysql 5.8K Aug 20 13:36 proclist_20260820_133650.txt [19:34:47] [7/7] > -rw-r--r-- 1 root mysql 1.8K Aug 20 13:37 proclist_20260820_133750.txt [19:35:50] i saw quite a few selects for the search index [21:22:36] ||smutstone|| [21:37:28] Wasn’t supposed to be merged yet, I had meant just for it to be temp deployed to see if I worked or not [21:38:03] oh, sorry! [22:01:10] Yeah, gonna have to dig deeper maybe to figure out the correct format for what it asks for since I don't know the codebase [22:01:41] can prob revert that, gonna have to try and figure out the proper solution and make a new pr to try [22:03:57] doesn't really work or do anything anyways so if you rather you just fix it on top we can do that [22:10:56] I'm gonna go and make a Phorge task for this even though I'd like to have more ideas, to make it easier for people to see and less likely to be totally lost [22:19:52] I eat databases [22:19:56] badge idea not a personal statement btw [22:24:48] can be both [22:24:54] honestly good name [22:25:20] I made my own I eat prod for breakfast badge on WO phorge lmao [22:25:28] but databases are tasty too [22:25:34] okay now i have to check WO to check [22:29:02] there's only a few funny badgegs [22:29:32] if you have any good ideas for new ones plz add once i finish this task [22:30:09] I have some ideas [22:40:00] https://issue-tracker.miraheze.org/T15878 [22:40:19] please contribute ideas so i feel less bad about asking for a badge that would be granted to me [22:40:54] I eat databases is a good one, it's a better name than I eat prod but less inclusives [22:42:09] @zipppee @wikitideharej @thewwrnerdguy poke as just talking about to or involved in my yapping yesterday [22:58:54] pr made for this now [23:00:11] for ease [23:02:40] [1/2] https://issue-tracker.miraheze.org/T15283#312240 [23:02:41] [2/2] You can use this script to add the redirects back automatically in the same PR. [23:03:07] Otherwise there'll be a 24-hour cooldown before redirects are automatically added back. [23:08:27] Which will still be fine because the redirects have been in place for a long time and a 24-hour disappearance doesn't really matter. [23:12:40] [1/2] For https://issue-tracker.miraheze.org/T15205, how do we feel about organizing CSP entries in ManageWiki? Do we have users select bare URLs in the config or the name of the service? The former leads to a lot of entries (more than 200 URLs in the CSP right now and 100+ of them will probably end up in the ManageWiki setting), but gives fine-grained control. The latter will reduc [23:12:40] [2/2] e the number of settings but remove certain choices (e.g. Discord has to be in `connect-src`, `frame-src`, `img-src` at the same time). [23:12:51] not entirely sure what to do here [23:13:21] do you think it would be easier for you to generate the list and update the PR? [23:13:33] Yeah that probably works better [23:13:48] I could probably hack my way around it but you already did all the hard work there