[06:50:14] poking our local @reception123 to get https://github.com/miraheze/mw-config/pull/5701 merged [07:08:53] done 🙂 [07:29:41] ty [09:09:59] That's smart BlankEclair [09:10:10] RhinosF1: he has vowed to merge pull requests [09:10:47] > [Thursday, 18 July 2024] [21.06.40 GMT+10] feel free to ping me if patches are left unmerged for too long 🙂 [09:11:06] BlankEclair: I meant the patch [09:11:09] oh [09:11:22] i hate unnecessary javascript with a passion [09:11:29] oo same [09:11:35] also, me irl: https://www.youtube.com/watch?v=4OwaCknLhVw [09:11:37] discord being an exception [09:11:52] still hate it, just tolerate it more than I would others [09:11:54] BlankEclair: the normal approach is to do it on wiki which ends up flashing and is really bad for perf [09:12:07] huh, flashing? weird [09:12:11] Also can't work on special pages [09:12:31] BlankEclair: if you do it through common.js/css then it loads the original logo and the temp logo [09:12:43] it doesn't change the logo image, just the link [09:12:53] Oh ah [09:12:58] and only the security-sensitive special pages [09:12:58] Maybe it's so early for code review [09:19:48] JavaScript runs on Special Pages [09:19:56] Just not login, sign up or preferences [10:06:40] yeah, can't even get CA to complete my onboarding, it's getting pretty bad [10:23:47] i onboarded my new phone into all my apps manually lol [10:23:53] but yeah, even i am getting busier ^^; [10:28:32] on another note, how do I even yaml 😢 [10:28:53] Why [10:28:56] like dude, why not JSON 😢 [10:29:04] Fair [10:29:39] I basically am learning how to write YAML files right now, I've never had to really deal with this before [10:30:00] SSL? [10:30:22] ig... [11:34:39] @bluemoon0332 I see what you're doing now and I know why you hate yaml, and probably puppet soon [11:34:56] puppet uses yaml? [11:35:03] I'm waiting for the moment I merge this and get 10 million errors in bots171 [11:35:04] or is it because of the custom language [11:35:24] I just know that's what's going to happen [11:35:47] BlankEclair: it can load yaml easier than it can load json [11:35:52] oh lmfao [11:36:16] Things like ssl, users module, now pywikibot all load yaml and puppet loads it in [11:36:19] Puppet uses its own DSL [11:36:30] Dont think there's a json way of doing that [11:36:43] But you can use yaml for things you would generally use yaml for and puppet can understand [11:36:54] Yup [11:37:03] Puppet itself is just yuck [11:37:05] Salt stack is better ❤️ [11:38:26] puppet isnt' so much of a language as it is an incantation phrase book powered by souls devoured thousands of years ago and given form by the heartache of opsen everywhere [11:39:40] I need to setup some configuration management at work and I just don't want to have to touch puppet config [11:40:38] Use salt or chef [11:44:58] It's going to be a mostly stable environment tbh [11:45:10] So I'm debating whether to just do it as prebuilt ISOs [11:45:25] And assume we don't need full config management [11:45:48] Cause once we set it up it's probably going to be rarely changed [14:49:10] BlankEclair: any idea why my NUMBEROFOPENREQUESTS isn't working (https://meta.miraheze.org/wiki/User:Reception123/temp_stats at the bottom)? [14:49:21] https://github.com/miraheze/CreateWiki/commit/7f7f86de6b6c7890dfd650b2b0be4274982d1cab (and the CWMagic separately done per CA's instructions) [14:50:29] it was cache [14:50:31] i purged the page [14:51:00] oh, I didn't know that would affect a yet to be implemented magic word [14:51:10] I do have to admit I never fully understood cache, it's just one of those things for me [14:51:45] parser cache to be specific [14:51:55] i assume you edited the page before implementing the magic word [15:03:01] @reception123 which cache [15:03:32] But ye what Claire said [15:03:50] Adding a magic word won't auto cause a PC regeneration [15:04:04] There's like 10 different caches though [15:04:44] action=purge worked so i assume parser cache [15:04:44] yeah that's why they confuse me [15:04:46] Browser, Edge, Parser, Opcache, object, query, men [15:04:59] no way they cache men [15:05:05] Typo [15:05:08] Mem [15:05:32] hi, I hear you guys are using cloudflare now [15:05:40] If it affects the text on the page, it's probably parser cache [15:05:49] Hi @cookmeplox, yes, mostly [15:05:58] Reception sent me some screenshots of some of the analytics, and I think you may still be getting like 75-80% of your resource usage from bots [15:06:11] @cookmeplox yes and? [15:06:33] 75-80 is probably high [15:06:34] and, I am wondering if you'd like some help getting rid of the particularly bad ones [15:06:38] no [15:06:46] based on the ratio of different HTTP versions I think it's about 80% [15:06:53] I've blocked the ones causing a problem [15:08:17] real user traffic is about 80% HTTP/3, but you guys are like 80% HTTP/2 [15:08:52] so I think there's a couple bots that are using browser UAs that are consistently eating a huge percentage of your overall resources [15:10:24] HTTP/3 is 65% of 200 code responses [15:10:41] @cookmeplox [15:10:54] hmm okay, perhaps i was seeing a longer time period from before you blocked them [15:11:07] or maybe they're all getting 403'd already [15:11:13] usually they stop after a while [15:11:40] but anyway, depending on how much you want to rely on the cloudflare stuff being available forever, you may want to consider just replacing your varnish stuff with it [15:11:57] you can cache the html pages extremely well that way too [15:12:23] and because you're on enterprise, you have access to purge-by-cache-tag, which is a game changer for this use case [15:12:51] Yes we know [15:13:10] we're replacing varnish, yes [15:13:13] Varnish is not used for traffic migrated to CF [15:13:27] much to my dismay (no -X PURGE) [15:13:41] at all? how are you doing html caching then [15:14:01] Cloudflare [15:14:03] varnish is used for the non-cloudflare traffic [15:14:14] https://meta.miraheze.org/wiki/User:BlankEclair/Caching [15:14:15] are you doing a cache-vary on the session cookie or osmething [15:15:15] Cache is bypassed if a session cookie is present with both Varnish & CF [15:16:09] what's your method for deciding what to purge when something is edited? [15:17:10] Same as was purged with varnish just authenticated to CF API rather than an PURGE request [15:17:50] i'm mostly curious if you're including redirects [15:18:06] because those are notoriously annoying to purge, since they're 200s and not 301/302s [15:18:25] a session cookie? [15:18:30] Does it check if it’s valid? [15:18:37] No [15:18:38] CF cannot validate that [15:18:43] i hope not, since that would require an origin request [15:18:49] Go check the source [15:18:49] which is the whole thing that's trying to be avoided [15:19:20] Hm, so adding whatever to the cookie field would bypass CF and hit the us directly? [15:19:24] Or am I missing something [15:19:24] i did and it's not particularly easy to tell, which is why I'm asking you [15:19:39] Yes [15:19:46] yes, but there's a bunch of easier things you could do to bypass CF so it doesn't really make sense as an attack vector to harden against [15:19:46] just adding ?wadfwadwad also does the same trick [15:19:59] Touché [15:20:27] @pixldev also you will eventually get blocked if you try and cache bust [15:20:33] Basically saying the missing grate on the vent is a way to get in when they left the window open [15:20:33] At any useful rate [15:21:53] There are rate limits and security rules in place that attempt to stop cache busting [15:22:27] anyway, i bring the redirects thing up just because you may want to switch to purging by cache tag [15:22:50] Duly noted [15:22:58] We'll take it under advisement [16:15:56] Where is the code that purges from Cloudflare when a page is edited on MediaWiki? [16:16:11] Because iirc we only have MultiPurge and that doesnt [16:18:06] Something does it [16:18:24] And the only thing we have is MultiPurge cause it definitely gets done [16:18:29] Cause we had the mobile issue [19:47:14] We might need to consider another extension to do that [19:47:20] cc @cosmicalpha [19:50:18] Is MultiPurge still being a bitch [19:50:36] Yes [19:50:45] We might need to develop our own extension tbh [19:54:20] Is it really so far gone we can’t even fork it? [19:56:11] No [19:56:17] I suggested we do that a while back [20:30:59] I don't see any code in MultiPurge that actually purges a page when its edited, as I said [20:31:14] all i see is code that purges when someone does ?action=purge or uses Special:PurgeResources [20:32:32] to purge on article edit needs to utilise the `PageSaveComplete` hook [20:33:22] Does it register any other hooks [20:33:47] none that do what people think it does [20:34:13] (Probably better to just put it into MirahezeMagic since it will be one hook [20:34:32] How tf [20:34:52] its intended to purge `load.php` urls, as it says in the extension description [20:34:59] no idea where the idea that it purges on edit came from [20:35:39] Has CF just [20:35:46] Not been doing that [20:35:47] This [20:35:48] Whole [20:35:49] Time [20:45:15] ¯\_(ツ)_/¯ [20:50:16] Mobile hasn't [20:50:22] No one has moaned about desktop though