[08:51:01] ema: nice work at https://phabricator.wikimedia.org/T293879#7450855 :) [08:52:45] is it possible to know if there is a gerrit patch for a given file? [08:53:05] Krinkle: thanks! I wouldn't have expected to work on optimizing mtail scripts but it's quite fun :) [08:53:28] XioNoX: you can filter for it like https://gerrit.wikimedia.org/r/q/file:modules/admin/data/data.yaml [08:54:41] majavah: hahaha perfect, I needed the same link with s/admin/network/ [08:54:41] thanks [08:59:49] volans: say you were starting a new (wmf) python project from scratch, what would you recommend as important things to have? mypy is a given, for example. [09:19:15] f-strings, also a given (i don't need to support stretch \o/) [09:20:13] kormat: sorry finishing one thing, will get back to you in few mins [09:20:27] ok :) [09:21:11] I have made some changes on our databases that serve watchlist on frwiki, jawiki, ruwiki and wikitech, if you notice something (errors, or timeouts etc) please ping me [09:30:10] kormat: back, so I'd say it depends also on the size of the project. If you don't want to make lot of decisions and want more or less all you could just copy the spicerack base structure. As for code best practices f-strings for formatting and pathlib instead of os.path, @dataclass when makes sense [09:30:58] TIL of @dataclass [09:31:43] volans: you have.. pyproject.toml + setup.cfg + setup.py. 🤯 [09:32:31] pyproject.toml was added with the introduction of black+isort [09:32:38] setup.py is for making deb packages easily [09:33:01] and still the real entry point unless you want the fancy poetry or similar [09:33:09] but then up to you to make them work with deb packages [09:33:12] and dh_python [09:33:39] * kormat winces [09:33:54] we can probably consolidate pyproject.toml and setup.cfg, I need to take a look at some point [09:34:54] i see that dh-python doesn't support pyproject.toml yet, as you say. sigh. https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=984824 [09:35:03] yep [09:35:25] that's why I'm sticking to setup.py for another while [09:35:29] * kormat nods [09:35:42] but if you don't plan to have debs [09:35:49] i do, sadly [09:35:55] :) [09:40:04] volans: have you tried sphinx with .md instead of .rst? [09:40:31] yes that works but at the time pypi was not parsing md correctly, not sure now [09:40:45] ah ok. not an issue here, thankfully. [09:41:15] rst is probably more powerful, but in general md should work totally fine AFAI [09:41:19] *AFAIK [09:41:32] rst is a massive pain [09:41:52] "oh, you changed the length of a heading, now you have to change the number of ='s underneath it". whyyy. [09:43:13] true! [11:15:55] [repeating because of the netsplit] [11:16:15] Since apt 1.9.11 (practically only bullseye+) one could set DPkg::Lock::Timeout to wait for the lock to be released instead of failing. Seems useful for both debdeploy and puppet-triggered apt actions. The default seems also sane (infinite wait on TTY, 2m otherwise). [11:16:19] details at https://salsa.debian.org/apt-team/apt/-/commit/b71b6802fe64723117bce6d91495a8af98ac528a#a810eb037294fdd3babe32d01f2f6655040fc47e [11:16:22] I found it through https://blog.sinjakli.co.uk/2021/10/25/waiting-for-apt-locks-without-the-hacky-bash-scripts/ [12:29:07] Did anybody ever had weird "system just freezes" issues with a X1 Carbon 7th? I do experience them from time to time and when that occurs it usually happens a couple of times in a row (like 5-15min after reboot it freezes again) [12:30:04] Nothing in logs obviously, writes to disk no loger seem to happen, no network connections possible ... memtest passes without issues [12:31:06] are you running an up-to-date system/kernel? I had such freezes because of an i915 bug but it was months ago [12:31:49] bullseye on 5.10 [12:32:18] https://gitlab.freedesktop.org/drm/intel/-/issues/2905 was the bug I had [12:32:28] I also do have that since I got the machine...so multiple debian and kernel versions already :) [12:33:08] temperature related? [12:33:13] It's sporadic enough to not care for some time. But on days it happens more often its very frustrating [12:33:42] no relation to temperature as well. got a node_exporter running and metrics do not show anything obvious as well :/ [12:34:26] maybe try configuring netconsole to another host in your network or something, in case you catch some of the logs [12:34:36] at that point I would run another OS and see what happens, but it doesn't look good [12:36:59] netconsole might be worh a try. Different OS is kind of hard as this is my work machine and I would like to keep working :p [12:40:46] it could be a live Ubuntu instance, maybe it's close enough to run with your $HOME [12:50:35] I'll try to catch something via netconsole first I think. Seems to be less hassle [15:15:34] Bullseye with 5.10 on x1 7th gen as well, but no I don't have this issue. [15:15:59] What I do have in the last couple of days is fwupmgr not updating [15:17:12] Weeks more like it. It says there are updates to apply, stages them, but on reboot I just see grub again, not the firmware updater [15:21:05] Hm. That I don't have. Happily applied a thunderbold FW update today [15:21:56] obviously since I've enabled netconsole it did not happen again (apart from one last time while still in the process of configuring it :)) [16:58:49] Hello, any idea how would I get rid of "total" in https://grafana-rw.wikimedia.org/d/X-ltxLV7z/mentor-dashbobard?orgId=1 for the WIP panel (updateMenteeData per data type)? [17:52:25] urbanecm: try adding `exclude(total)`? [17:53:04] cwhite: thanks, works perfectly. [19:00:53] urbanecm: In the future (and as prep for new Metrics lib and Prometheus) I'd recommend sending those metrics under a different name, e.g. growthExp_menteedata_total and growthExp_menteedata_db.* [19:02:22] or e.g. _total.$db and _type.$db.$type [19:03:43] though in this case, since total can be derived, with a sum() might not need that at all actually