samueldr changed the topic of #nixos-infra to: NixOS infrastructure | logs: https://logs.nix.samueldr.com/nixos-infra/
cole-h has quit [Ping timeout: 268 seconds]
cole-h has joined #nixos-infra
cole-h has quit [Ping timeout: 252 seconds]
thefloweringash has quit [Ping timeout: 276 seconds]
thefloweringash has joined #nixos-infra
supersandro2000 has quit [Killed (verne.freenode.net (Nickname regained by services))]
supersandro2000 has joined #nixos-infra
rtjure has joined #nixos-infra
cole-h has joined #nixos-infra
<gchristensen> ikwildrpepper: how many more macs can we fit in NL1 before the datacenter's remote hands gets in to hot water?
<ikwildrpepper> gchristensen: dunno, would have to check if we need more shelves then
<ikwildrpepper> I think we have enough internal power plugs and network ports
<gchristensen> when do we get NL1 a separate measured circuit for power
<ikwildrpepper> it uses about 200-250W
<gchristensen> https://nix.ci/prometheus/graph?g0.expr=ofborg_queue_builder_waiting&g0.tab=0&g0.stacked=0&g0.range_input=12h we've opened up ofborg to all PRs by default and predictably the one machine isn't keeping up exactly :P
<ikwildrpepper> figures
<gchristensen> I tried to find a refurbished mac mini here in town but the local apple shop (not an applestore) doesn't have any used, and doesn't have any new either
<andi-> Who has permissions to write to tarballs.nixos.org? https://github.com/NixOS/nixpkgs/pull/105026#issuecomment-843345185
<gchristensen> I can get you
<gchristensen> I think I had a script for this
<gchristensen> but maybe terraform would work too
<thefloweringash> Thanks!
<gchristensen> something is funny with the hashes, I'm not sure why we're getting different hashes
<gchristensen> d'oh :)
<gchristensen> so, uh, let's merge? https://github.com/NixOS/nixpkgs/pull/123557
<gchristensen> I'll go ahead and do it unless someone else wants to have the honor
<andi-> gchristensen: do it
<gchristensen> andi-: before hand we should run through https://github.com/NixOS/rfcs/blob/master/rfcs/0046-platform-support-tiers.md
<gchristensen> reading through it now
<gchristensen> seems like tier 3?
<MichaelRaskin> gchristensen: sounds like that
<andi-> >Native bootstrap tools available
<andi-> well while that is true those don't work :D
<andi-> Does that qualify?
<andi-> That document also assumes MacOS is always Tier-2..
<MichaelRaskin> That document says the list needs amending if situation changes significantly!
<gchristensen> is there a way to fan out to aarch64-darwin without requiring it for advancement?
<MichaelRaskin> You mean at Hydra?
<gchristensen> yeah
<MichaelRaskin> Sigh, how much delay is OK?
<andi-> don't add it to supported platforms and add a ne wjob?
<andi-> *new job
<gchristensen> so, really, this "supportedSystems" and "limitedSupportedSystems" should be replaced with tiers
<MichaelRaskin> Everything needs replacement with grayscale-aware sliding scales. Everything.
<gchristensen> :)
<MichaelRaskin> I mean, you could have a separate jobset for aarch64-darwin, or you could even have it take the _channel_ branch only
<gchristensen> I think the thing to do for today is make it supportedSystem for the -darwin channel, not for the main nixpkgs-unstable channel
<andi-> IMO since this architecture is still very young it shouldn't block the normal channels and also likely needs more world rebuilds in the short term
<andi-> MichaelRaskin: as you are the author of that RFC: In Appendix A, Tier 6 it says "ongoing to merge Tier 4" shouldn't that be Tier 5"?
<MichaelRaskin> Ooops, in the definitions it was indeed fixed
<andi-> Or was the intention at that time to target tier4 and skip 5?
<MichaelRaskin> In a sense, nobody truly targets 5…
<MichaelRaskin> But no, the real story is that Tier 2 was originally 1.5, and i686-linux was 2-ε
<andi-> Somehow a platform should probably enter at the bottom and then (eventually very quickly) reach 1-3, no?
<MichaelRaskin> As usual, all this RFC was basically written for a single line of it
<MichaelRaskin> > Platform support status at the Tier 4 and below is updated according to achievement of the requirements defined in this RFC.
<MichaelRaskin> BTW, has all the manual rehashing converged? Because if so, the list could be updated with the recent work and finally pasted as an appendix to Nixpkgs manual…
<MichaelRaskin> But basically RFC intentionally does not restrict aarch64-darwin suddenly being declared Tier 3 as long as the checklist is there
<MichaelRaskin> Tier 3 is not a guaranteed thing, as it is as resource commitment, but of course aarch64-darwin will get whatever is expected from the Hydra side.
<gchristensen> maybe gently
<sterni> the infamous temprorary measure of 5 years ago
<sterni> I think some of it has been reverted already
<gchristensen> I'm a firm -1 on reverting that commit and adding aarch64-darwin in the same month
<sterni> for the jobset perl and ocaml remain
<sterni> python has been reenabled
<sterni> I guess I'll reenable ocaml when I remember in the next weeks, I want to get a better overview over its state
NinjaTrappeur has quit [Ping timeout: 250 seconds]
NinjaTrappeur has joined #nixos-infra
supersandro2000 is now known as Guest2135
Guest2135 has quit [Killed (orwell.freenode.net (Nickname regained by services))]
supersandro2000 has joined #nixos-infra
<sterni> any ideas if there could be another explanation than just GHC things that ghcHEAD on aarch64 times out after 10h on hydra?
<gchristensen> is it big-parallel?
<sterni> gchristensen: no, but so are none of the other GHC derivations
<sterni> it's hard to judge because I don't really have any data regarding GHC on aarch64
<sterni> on x86_64-linux compilation times are pretty reasonable usually like < 1h on pretty old hardware
<gchristensen> is it actually GHC, or is it something built with GHC?
<sterni> gchristensen: GHC unstable https://hydra.nixos.org/build/142788116
<sterni> oh wait wrong link
<gchristensen> I wonder if it is using too many cores for compilation again
<gchristensen> https://gitlab.haskell.org/ghc/ghc/-/issues/9221 -- (super!) linear slowdown of parallel builds on 40 core machine
<sterni> oh wow there is still one running https://hydra.nixos.org/build/143298227/nixlog/2/tail
<sterni> it is even doing stuff
roberth has quit [Ping timeout: 258 seconds]
zimbatm[m] has quit [Ping timeout: 258 seconds]
thefloweringash has quit [Ping timeout: 245 seconds]
roberth has joined #nixos-infra
zimbatm[m] has joined #nixos-infra
thefloweringash has joined #nixos-infra