20 Oct 2024 |
hexa (signing key rotation when) | so they'll slowly migrate to 2.24 over the next few days | 00:16:46 |
Mic92 | hexa (signing key rotation when): thanks. I think John Ericson added some better error handling for parsing machine definitions to nix, because I think he faced a similar issue? Can you give me the failing machine definition? | 11:19:25 |
hexa (signing key rotation when) | e.g. root@2d838f2f.packethost.net x86_64-linux,i686-linux /var/lib/hydra/queue-runner/.ssh/id_ed25519 24 1 kvm,nixos-test - c3NoLWVkMjU1MTkgQUFBQUMzTnphQzFsWkRJMU5URTVBQUFBSU8yWXpEZzdHZ3Y5K3lDNVhCeWlNcW1wbE5WTFhwZFdKcUhTc0tRYXpJU0Q= | 12:32:12 |
Mic92 | We were thinking about setting up a staging hydra to test the upcoming s3 cleanup (my plan b). I also want to use it to test patches that you were pointing out. | 12:56:04 |
John Ericson | Yeah maybe what happened was when I started using Nix's machine file parsing things got decoded twice | 14:13:30 |
John Ericson | Ah I see you all found it already, good! | 14:44:02 |
John Ericson | Let's get that fix on the `nix-2.21` branch and then merge it into each successive `nix-*` branch and master. | 14:46:16 |
Mic92 | John Ericson: I would like to document this branching model that you use for hydra. How does it work? And what nix branches do we actually want to support? | 16:25:57 |
emily | given that Nix only support 2.24 and (for a few more months) 2.18 it seems pointless to go further for Hydra | 16:27:08 |
John Ericson | @joerg:thalheim.io: (terse cause phone) basically instead of back ports I do merges | 16:27:19 |
John Ericson | You can re-merge older nix-* into new nix-* all you want | 16:27:54 |
John Ericson | master uses the latest nix release | 16:28:09 |
John Ericson | nix-next uses Nix master | 16:28:18 |
John Ericson | nix-* has "the newest stuff" as long as it uses the right nix version | 16:29:43 |
John Ericson | So sometimes it's convenient to make a change in an older nix-* and then merge merge merge until on reaches master | 16:30:45 |
hexa (signing key rotation when) | Mic92: how does the devshell work in the terraform dir? | 23:22:09 |
hexa (signing key rotation when) | I get neither aws nor terraform even though I have opted into direnv | 23:22:20 |
hexa (signing key rotation when) | hm ok, it seemingly got confused by the nested devenvs | 23:29:19 |
21 Oct 2024 |
nh2 | WHo manages the AWS AMIs?
This one is missing already for 6 months with no reply:
https://github.com/NixOS/nixpkgs/issues/307519 | 02:10:28 |
emily | Arian might know | 02:11:24 |
nh2 | * Who manages the AWS AMIs?
This one is missing already for 6 months with no reply:
https://github.com/NixOS/nixpkgs/issues/307519 | 03:35:45 |
nh2 | In reply to @emilazy:matrix.org Arian might know I'd definitely appreciate some insight! | 03:42:24 |
Mic92 | @nh2:matrix.org: it's @arianvp:matrix.org there is also an aws nix matrix channel | 03:59:43 |
Mic92 | @hexa:lossy.network: if you only use direnv, than it should unload one devshell before it enters a new one | 04:01:35 |
Arian | I replied to and closed the issue | 07:22:55 |
Arian | The website has been fixed since 24.05 already. Can't reproduce. | 07:23:26 |
sterni | vcunat: what's the darwin queue like these days? haskell-updates is going to be ready in the next few days and I wouldn't necessarily want to merge 14k darwin rebuilds into master if I can avoid it. | 11:06:36 |
toonn | Looks like currently there's almost 90k builds queued for both Darwins. Don't know about the rate of decline currently though but they're over 30k builds behind Linux. | 11:15:39 |
vcunat | In reply to @sternenseemann:systemli.org vcunat: what's the darwin queue like these days? haskell-updates is going to be ready in the next few days and I wouldn't necessarily want to merge 14k darwin rebuilds into master if I can avoid it. Perhaps merge to staging-next early (say, today?) and do potential fixups in there during this week? | 11:50:52 |
sterni | that sounds like an idea, could do it this evening | 12:08:13 |