gchristensen changed the topic of #nixos-borg to: https://www.patreon.com/ofborg https://monitoring.nix.ci/dashboard/db/ofborg?refresh=10s&orgId=1&from=now-1h&to=now "I get to skip reviewing the PHP code and just wait until it is rewritten in something sane, like POSIX shell. || https://logs.nix.samueldr.com/nixos-borg
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-borg
hmpffff has joined #nixos-borg
hmpffff_ has quit [Ping timeout: 272 seconds]
evanjs has quit [Ping timeout: 244 seconds]
evanjs has joined #nixos-borg
cole-h has quit [Quit: Goodbye]
orivej has quit [Ping timeout: 265 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
hmpffff_ has joined #nixos-borg
hmpffff has quit [Ping timeout: 244 seconds]
hmpffff has joined #nixos-borg
hmpffff_ has quit [Ping timeout: 260 seconds]
<LnL> manually killing staging build is going to become annoying real soon...
infinisil has quit [Ping timeout: 246 seconds]
infinisil has joined #nixos-borg
<gchristensen> :/
sphalerite has quit [Quit: rebooooooooot]
sphalerite has joined #nixos-borg
sphalerite has quit [Client Quit]
sphalerite has joined #nixos-borg
<LnL> look at that i made it through :)
<gchristensen> :D
<LnL> how confident are you in nix/nixpkgs-update.sh for infra?
<LnL> I think that would fix the log levels
<gchristensen> (in a meeting, ping in 1h?)
cole-h has joined #nixos-borg
orivej has joined #nixos-borg
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-borg
<{^_^}> ofborg/infrastructure#18 (by LnL7, 34 seconds ago, open): nixpkgs: db3e832 -> a7ceb25
<cole-h> For colored promtail, yes? (Or, at least hoping this makes colors work like they did for you locally.)
<LnL> no idea what else it could be
<LnL> also I hope it's actually just loki and not promtail, if so it will most likely work retroactively
<gchristensen> LnL: what branch is that update from?
<LnL> still 19.09
<gchristensen> cool, should be fine
<LnL> seems like somebody must have backported a loki update
<LnL> this is still most likely going to restart everything tho, no issues with that?
<gchristensen> let's find out =)
<cole-h> Haha
<LnL> alright
<cole-h> Worst case, we redeploy HEAD^, right?
<cole-h> (Well, bad case)
<cole-h> (Worst case, the world ends)
<LnL> by the way since I've been messing with my desktop so much lately I finally made the setup a bit nicer
<LnL> deploys with a git push now
<LnL> I think there might be a neat idea there somewhere
<gchristensen> nice! :D
<cole-h> desktop -> darwin machine?
<gchristensen> sounds good
<LnL> both local push and remote
<LnL> so I'm forced to actually commit my stuff
<cole-h> lol :D
<cole-h> Oh wow, that's really cool
<cole-h> Do I smell a blog post coming on? :eyes:
<gchristensen> WOW
<gchristensen> I hope so LnL :D
<cole-h> Pretty please?
<LnL> I should make a blog first then
<cole-h> You could use a `gist` for that, temporarily ;)
<LnL> hehe
<LnL> not entirely clear yet how I want to deal with channels, etc.
<cole-h> Just don't. Make the jump to flakes :D
<gchristensen> lol
<LnL> I've totally not looked at that yet
<LnL> know next to nothing about flakes
<cole-h> Same
<cole-h> I just know it has this weird `nixpkgs#bashInteractive` syntax
<LnL> Exception: activation of 3 of 3 machines failed (namely on ‘packet-spot-eval-3’, ‘packet-spot-eval-2’, ‘packet-spot-eval-1’)
<cole-h> Welp
<LnL> sad-noises.wav
<LnL> oh "error: experimental Nix feature 'nix-command' is disabled"
<cole-h> If deploy fails, does it auto rollback?
<LnL> if deploy fails it didn't deploy :)
<LnL> I hope...
<cole-h> It says "restarting the following units" so I didn't know if that was "if this works, these will be restarted" or "these are being restarted"
<LnL> euh.. this is pretty bbad
<cole-h> Wait
<cole-h> I see colors
<cole-h> Nani?
<LnL> 何?
<LnL> yeah I think it mostly activated, just returned 1
<cole-h> (actually 2)
<cole-h> (well, 1 and 2)
<cole-h> I think it sucks that if you have an unknown experimental command, it gets treated as an error
<LnL> no that's fine but the nix commands where used for important things before marked as experimental
<cole-h> LnL++ The presence of colors makes this much easier to skim. Thanks!
<{^_^}> LnL's karma got increased to 51
<LnL> gchristensen: nixUnstable is kind of a leftover from the 0.12 days right?
<gchristensen> yeah
<cole-h> 0.12? Wow, that's a long time ago
<LnL> has it been that long?
<LnL> 0.12 == 1.0 btw
<LnL> that's better :)
<cole-h> Still, 1.0 is a long way from 2.3.4 :P
<LnL> oh no I meant 1.12
<LnL> and 2.0 respectively
* LnL was thinking of versions from another project
<cole-h> Keep confusing me why don'tcha
<cole-h> lol
orivej has quit [Ping timeout: 272 seconds]
<cole-h> Hahaha
<cole-h> We have both `phpfpm.nix` and `php-fpm.nix` in the infra repo lololol
<cole-h> Will replacing `/run/php-fpm.sock` with `${super.config.services.phpfpm.main.socket}` work?
<cole-h> Sorry, `super.config.services.phpfpm.pools.main.socket`
<LnL> look for symlink compat for another fun/weird thing
<cole-h> ?
<cole-h> Heh, I remember seeing that.
<{^_^}> ofborg/infrastructure#20 (by cole-h, 11 seconds ago, open): No more warnings
<cole-h> Since `phpfpm.pools.main.listen` gets removed in this, some extra caution is warranted (e.g. the secrets/encrypted stuff should be checked for references to that as well)
<cole-h> Also, if we were relying on the silent concatenation of `types.string`
<LnL> very nice :)
<{^_^}> [ofborg] @LnL7 opened pull request #481 → skip stdenv rebuilds → https://git.io/JfZOF
<cole-h> LnL: That's to help combat the binutils/coreutils thing we ran into recently, right?
<cole-h> (eval failing because one step was actually a build)
<LnL> ah no
<LnL> just not to start staging builds so they don't hold up the lonely darwin builder
<cole-h> Oh, got it
* cole-h imagines the darwin builder sweating hard every time a staging PR comes in
<LnL> 86973 was the last one causing most of the queue you see now
<LnL> and I spy Skip: 'flibgsf'
<LnL> as well as some builds that still did something
<LnL> ideally this would just tigger a requeue with TAGS but that's a bigger thing
<cole-h> TAGS?
<LnL> let's see if I can find that link somewhere
<cole-h> lol. My Google-fu usually fails me on these acronyms-shadowing-commonplace-items
<cole-h> Also, how does this determine if it's targeting staging? Is that related to `nix-store -q --size`?
<LnL> hopefully that'll make it easier to find next time
<cole-h> "Task Assignment based on Guessing Size" interesting
* cole-h files it away with the growing pile of academic papers to read eventually
<LnL> so some context for the stdenv check
<LnL> the first two steps are just to get the store path of the stdenv (for the current pr)
<LnL> a bit awkward but there's not really a nice way with the nix-* commands
<LnL> then the last queries that store path, but instead of on the local machine it asks the cache
<LnL> --size is just because that's one the simplest query commands exposed
<cole-h> Huh
<LnL> eg. nix path-info /nix/store/8z47adw4zaw9q8vybw8r1wp2rg9r0r7r-hello-2.10 most likely fails with path is not valid since you don't have that in your local store
<cole-h> So first command gets you /nix/store/stdenv.drv, second command gets you /nix/store/stdenv, and third command checks if that stdenv if cached? Did I get the gist of it?
<LnL> but nix path-info /nix/store/8z47adw4zaw9q8vybw8r1wp2rg9r0r7r-hello-2.10 --store https://cache.nixos.org will return because it's a valid path in the cache
<LnL> yep
<cole-h> I think that should be documented in-line, if possible :^)
<cole-h> Or as a doc comment for the function
<LnL> yeah, good point
<LnL> with the new commands this would be so much nicer: nix path-info -f. hello --store https://cache.nixos.org
<cole-h> Hopefully they stabilize soon-ish :^)
<LnL> it's like using git-hash-object & friends to make a commit :p
<cole-h> lol :D
<LnL> anyway, trying it out so hopefully I won't wake up to a giant gueue again
<cole-h> That thing was monstrous
<cole-h> Looked at the dash this morning, saw it reached about -60 at one point :D
<LnL> yeah and unlike before they won't magically disappear with a deploy anymore
<cole-h> Looks like those 2 tests fail because it skips the stuff that doesn't exist in the cache lmao
<cole-h> and it doesn't output what the test wants
<LnL> yeah needs a bit more work
<cole-h> I always get bamboozled in my ofborg checkout. I try to `git checkout master`, which checks out master from *your* fork (since I added you as a remote)
<cole-h> and then I remember our master is actually "released"
<gchristensen> hah
<gchristensen> yeah
<cole-h> It happens roughly once every time I touch ofborg lol
<cole-h> gchristensen: I'm thinking of updating the README to remove mention of "known users" and similar old stuff. I was thinking of expanding it to also note that `@ofborg` works the same as `@GrahamcOfBorg` -- yay/nay?
<gchristensen> doit! :)
* cole-h also changes the headline from `# grahamcofborg` to `# ofborg` :^)
<gchristensen> yay
* cole-h also references @ofborg before @GrahamcOfBorg in all applicable instances
<cole-h> ;^)
<gchristensen> <3
<cole-h> gchristensen: Should `config.{extra-,}known-users.json` be removed?
<gchristensen> sure!
<gchristensen> might need to be some fixups in the infra repo
<cole-h> Oh, good point.
<cole-h> Seems to only be referenced in `config_private` -- simple removal there
<cole-h> gchristensen: "authorized users" == "trusted users"?
<gchristensen> uhh not sure
<gchristensen> I need more context :)
<gchristensen> yeah
<cole-h> Cool, will reword that, too, then.
<gchristensen> cole-h++
<{^_^}> cole-h's karma got increased to 41
<cole-h> Also, does the "old php stuff" need to be kept around? AFAICT, it's present in the infra repo
<gchristensen> which? probably yes
<cole-h> The "Only Graham needs to do this, since I run the only remaining PHP components"
<gchristensen> oh delete
<cole-h> The contents of which are in infra/nixops/modules/webhook/config.php
<cole-h> OK yeah, figured
<cole-h> Huh, who knew
<cole-h> You can have commas in git branch names
<cole-h> gchristensen: You're gonna love the branch name I picked for this (once I push it out)
<gchristensen> :D
<cole-h> (Note: That will probably be later tonight to tomorrow morning due to lecture starting soon.)
<gchristensen> :D
<gchristensen> learn well
<cole-h> Fingers crossed my VM doesn't freeze (but only some parts of video rendering -- could still click some stuff and audio worked fine???)
orivej has joined #nixos-borg