[14:44:20] Reedy: I'm back from vacation and reviewed your changes to the token patch, I think we should merge it. Not sure who of us should +2 though [14:45:08] Pchelolo: If you want to +1 it.. I can then +2 it. Or we can ask someone like James_F nicely to do it if we want a third party [14:45:19] +1d [14:45:30] it depends on a trivial wikibase patch as well [14:53:41] Yeah, +2 ing that isn't an issue [16:02:02] * James_F query Reedy [16:02:10] Ha. That didn't work. :-) [16:40:40] Thanks James_F! :) [16:40:54] Happy to help. [19:54:46] Krinkle: is it possible to profile a request in production and get a flame graph of it? or is the data available in xhgui the best i can get? i think i recall you talking about this recently somewhere. [20:13:58] MatmaRex: that's what we offer today indeed but you can do it manually by hacking an mwdebug server and use excimer instead tideways with a short sampler interval, write to a tmp file, and feed to flamegraph.pl in your home dir or something. [20:16:23] okay, thanks (i don't have production access so that's not a practical option) [20:18:36] MatmaRex for root^W^W^W^Wshell access [20:18:48] D: [21:22:53] Pchelolo: around? [21:22:58] yup [21:23:33] Pchelolo: https://github.com/wikimedia/limitation/pull/6#issuecomment-918502277 [21:26:23] oh.. that stuff... [21:26:57] Pchelolo: yes it broke my server [21:28:17] honestly it's far past due to get away from all this stuff... [21:29:15] Pchelolo: is there anyway we could unbreak it now though so I'm not left with a broken bullseye VM [21:30:03] what do you have installed and what exactly is breaking.. if we just revert this, we'd end up breaking whatever Kosta was fixing there... [21:30:06] RhinosF1: that's actually a bug in npm [21:30:12] npmv7 specifically [21:30:32] Pchelolo: it's defaulting to pulling over ssh [21:30:50] legoktm: Any ideas [21:31:04] https://github.com/npm/cli/issues/2610 [21:31:10] stay on npmv6 until they fix it? [21:31:45] legoktm: bullseye [21:31:57] Are you telling me I need to get the package from not Debian [21:32:37] well I don't know the broader circumstances of what exactly you're doing [21:32:56] but there's a reason we haven't moved to npm v7 yet, and that's (partly) why [21:33:21] legoktm: we reimaged our first server of the prod ones to Debian bullseye and so far it's fell flat on our face [21:33:33] Somehow test didn't pick up a lot of issues [21:33:54] I certainly wouldn't run npm on a prod server [21:34:05] We started with https://gerrit.wikimedia.org/r/c/3d2png/+/720699 [21:34:13] Which paladox tried to cherry pick [21:34:13] In theory switching to git+https://github.com/wikimedia/kad ... would avoid the issue entirely [21:34:29] And eventually ended up with ssh issues because that patch I linked above [21:35:04] git+https seem to be breaking our CI [21:35:17] I guess we could also publish the kad fork under something like @wikimedia/kad-fork and just depend on it like any other regular package [21:35:30] that will definitely avoid both problems... [21:36:27] that seems like an even better option [21:36:54] Pchelolo: how long does that take [21:37:21] not too long. lemme try that [21:37:34] paladox: ^ [21:38:19] thanks! [21:38:27] Filed https://phabricator.miraheze.org/T8032 RhinosF1 [21:38:57] paladox: ok ty [21:39:05] I need to sleep [21:39:19] Because I'm up early tommorow and stupid cold already making me drowsy enough [21:39:28] Well I say cold I mean paracetamol [21:40:11] I'll poke at this a bit later today [21:41:12] Cool