[08:47:34] hashar: hello ! do you have any eta on https://gerrit.wikimedia.org/r/c/operations/software/netbox-deploy/+/1004192 ? It's blocking us from moving forward with some Netbox related projets [09:08:08] XioNoX: looks like I have missed volans reply from monday:-( [09:09:07] I guess the easy fix is to pin the python build image to the last known version before the one that did `USER nobody` [09:09:18] aka run everything as `root` [09:19:52] hashar: it's not critical urgent, but just want to make sure it's on someone's radar. It for example could prevent an emergency deployment (but we never had to do those so far) [09:30:48] XioNoX: I don't get it it :) [09:30:58] it is blocking you but at the same time it is not urgent? ;D [09:31:06] anyway, I don't know anything about those deployment/freezing scripts [09:31:11] but I can look at making them work [09:31:34] it is probably not too complicated but I gotta dig [09:40:36] hashar: for the last comment I put there, I haven't digged but it looked weird it added virtualenv as a dep and also wtith that so specific version, I'm not sure how that's related [09:40:47] I think I got it [09:40:52] I have used `pip install --target` [09:40:52] unless is a by-product of the warnings and sed error in the previous step [09:41:04] which as --prefix or --root is broken in subtile ways or not doing what I would expect [09:41:11] and I got hit by that in another project [09:41:20] I am repurposing the patch to use a virtualenv [09:41:26] hashar: blocking, important, not urgent :) [09:41:49] and better not wait it becomes urgent ;) [09:41:50] the netbox plugin is installed in the same dir as netbox itself and "pip install -t" complains cause it isnot meant to install multiple packages in parallel as I got it [09:41:59] XioNoX: yeah I got it :) [09:57:47] volans: XioNoX: I have made some fix up with patchset 4, notably using a virtualenv to install netbox and the plugin instead of the stupid `pip install --prefix` [09:57:48] https://gerrit.wikimedia.org/r/c/operations/software/netbox-deploy/+/1004192/3..4 [09:57:49] is the diff [09:57:55] and locally that seems to work without errors [09:59:06] hashar: thnks, I'll test it shortly [09:59:08] <3 [10:21:38] claime, topranks, do you have a host handy I could use to test https://gerrit.wikimedia.org/r/c/operations/cookbooks/+/981472 ? [10:22:35] XioNoX: sretest2003 I think can be used [10:22:46] let me double check what state it’s in [10:25:19] XioNoX: yeah sretest2003 is on private1-a-codfw in rack a8 so probably perfect [10:25:38] nice ! thx [10:25:50] thank you! [11:18:34] volans: the last sed issue is actually a bug in `sed` !! :) [11:18:46] but if it does cause any troubles, I guess it can be ignored [11:19:08] permission denied is a bug in sed? :D [11:19:34] https://gerrit.wikimedia.org/r/c/operations/software/netbox-deploy/+/1004192/comments/642b5c97_6ad812f4 [11:19:35] yeah [11:19:37] :D [11:20:03] > Previously sed would incorrectly set umask on temporary files, resulting in problems under certain fuse-like file systems [11:20:11] ahhh nice finding [11:20:14] sure, ignor eit [11:20:17] and I think I remember you use a Mac [11:20:23] so I guess Docker desktop goes into the way [11:20:26] and again: [11:20:30] yep [11:20:31] Stop using a Maco :) [11:20:33] sx [11:20:34] ahahah [11:20:35] ahah [11:20:36] * hashar giggles [11:20:41] XioNoX: I think we can merge it [11:20:42] it is just too much troubles down the line [11:20:56] or a great excuse to not use docker :D [11:21:02] I have used to have a Macbook Air and definitely loved a lot of it [11:21:36] then in the end I installed much softwares with homebrew, used terminals and firefox, so I had not much advantages using mac os x over Debian [11:22:08] I swapped, and well my overall experience is much better (although I do miss the nice Apple UI, I am not regretting the broken packages or proprietary annoyances) [11:22:13] end of my rant [11:22:14] :) [11:22:24] so yeah that should do it [11:22:50] for netbox-deploy, ideally I'd love to stop relying on a huge artifacts.tar.gz of doom and instead move toward tracking wheels individually [11:22:51] yeah everything has pro and cons, and the new cpus are basically not usable for us [11:23:28] but that is not as straight forward as one can imagine. The alternative is to find a way to make python installs reproducible :/ [11:23:53] volans: I think that's changed with the improvements to rosetta now [11:24:21] Still slower than x86 for x86 images, sure, but not 1h30 to run puppet tests [11:24:47] (I also find the Macs to be grossly overpowered for my use case, given my Lenovo is from 2018 and still feels super responsive to me) [11:25:03] anyway. Lunch & [11:25:10] claime: did the fix it? On Sonoma when it went out docker couldn't run at all [11:25:33] I run on ancient hardware at the moment :d [11:25:38] volans: I don't have first hand experience, I returned my M1, but brouberol has a newer macbook [11:25:47] I'll ask thx [11:32:29] hashar: as for the repo, we could do single wheels, that's not a big deal, we do untar it on deploy [11:34:03] we can also delete the file, force it's deletion from the repository archive and then re-create it [11:49:35] hashar: if your computer feels overpowered, can I suggest you try switching to emacs? ;-) [12:24:16] I have a m3 for WMF (and have run our puppet tests on my personal m2 in the past). There was a point in Docker where running a single test went from 20s to 2s, and I saw a "performance improvement for apple sillicon" in their changelog [12:25:00] which feels good enough TBH. Obviously not what you get w/o emulation at all, but still good enough [12:40:24] o/ [12:42:32] volans: we have some workflow for tracking individual wheels in integration/zuul/deploy, I 'd have to describe it somewhere and/or upstream the process to the python build images. It is a bit tedious though :) [12:43:14] brouberol: is your virtual terminal GPU accelerated? =) [12:44:41] elukey: Welcome back! [12:44:49] thanks! <3 [12:44:56] reading emails atm :D [12:45:21] I'm not even sure TBH. I use stock iTerm2. The only change I make is to the font [12:45:25] welcome back elukey! [12:46:22] as time goes on, I'm less and less interested in tweaking my own personal computer. I just care that it works, so I don't have to spend more time computering than I already do [12:47:31] so much so that I just realized that the question might have been a troll (or not), and at that point I just don;t know ^^ [15:53:20] <_joe_> brouberol: same for me, I have enough interesting computer problems that are related to work and thus more relevant than my personal device [15:59:10] One of my friends is a real "desktop productivity maximizer"...I mainly steal his recommendations. Or sometimes when I'm pairing with one of y'all I'll see something cool [15:59:34] Otherwise I'm too busy or stubborn to change much on the desktop [16:03:27] <_joe_> I will soon have to do the jump to wayland and I dread that some things will not work like they used to. [16:03:53] * Emperor hugs their fvwm setup [16:53:59] _joe_: FWIW it took me way way way less time to configure sway than i3/x11 - the latter of which was a constant battle due to the bugginess. I've been more-or-less running the same config since 2018 [18:09:48] I see "class passwords::grafana::production" with a comment # Deprecated 2017-01-18 in the private repo, cleaning up some stuff. I see nothing in public puppet that includes passwords::grafana. this is sane to delete, right :) [18:10:13] (also others like racktables already done.. and whatnot) [22:02:06] inflatador: thanks for picking up T358727 as a redirected request from Traffic. please let us know if we can be helpful :) [22:02:07] T358727: Reclaim recently-decommed CP host for WDQS (see T352253) - https://phabricator.wikimedia.org/T358727