Sender | Message | Time |
---|---|---|
4 Feb 2022 | ||
In reply to @buckley310:matrix.orgWhat I dislike about that approach is that it feels like you'd lose the flake advantage, as it makes it easy for the colmena definition to diverge from the flake one | 18:51:34 | |
well its what ill go with for now. its easy enough to not diverge, granted im the only cook in the kitchen on this one | 18:53:48 | |
i dont know of a better way at this moment | 18:54:30 | |
Yes, of course. I'd just be worried that one of the evals behaves slightly different somehow | 18:54:36 | |
they behaved slightly differently before :) | 18:55:26 | |
probably due to the impure build or something, they would produce different store paths for the systems than nixos-rebuild --flake would | 18:56:08 | |
hey team i'm hoping someone can help me out here... here is my
i'm running into a mountain of errors when i run | 20:32:36 | |
here is a snippet:
| 20:33:05 | |
for reference
| 20:33:55 | |
In reply to @aanderse:nixos.devThere was a bug that got fixed in 0.2.1 as well as in main: https://github.com/zhaofengli/colmena/issues/50 | 20:34:13 | |
It occurs in nixos-unstable with sandboxed documentation builds enabled, where documentation.nix uses modulesPath to strip off the prefixes of module paths | 20:35:20 | |
should i backport 0.2.1 to nixpkgs stable for you? 😄 | 20:35:32 | |
oh | 20:35:58 | |
nevermind... i see now | 20:36:02 | |
ha | 20:36:03 | |
thank you so much Zhaofeng Li ! | 20:36:18 | |
i'm too lazy to migrate from morph 😓 | 20:37:10 | |
Colmena should hopefully just work with your existing morph configs | 20:38:43 | |
you do have to change how you pin nixpkgs, though | 20:39:05 | |
* you do have to change how you pin nixpkgs, though: https://zhaofengli.github.io/colmena/unstable/tutorial/migration.html | 20:39:27 | |
hexa: i haven't used morph ... but i can't say enough good things about colmena this project is a breath of fresh air after having to deal with nixops issues (no stable release on python3 ) for a while | 20:39:33 | |
well … nixops is on another level 😀 | 20:40:14 | |
i don't mean to overwhelm Zhaofeng Li with compliments... but what a wonderfully simple tool to use ❤️ | 20:40:16 | |
morph just seems to be … enough for the maintainers | 20:40:23 | |
In reply to @hexa:lossy.networkYes indeed, NixOps does much more than morph/colmena aims to do with the IaaS features | 20:43:20 | |
i moved from morph. its missing some creature comforts 🤔 | 20:41:48 | |
People have been asking about provisioning in Colmena, and that's better handled by tools that are focused on provisioning (e.g., terraform) than Colmena itself: https://github.com/zhaofengli/colmena/issues/42 | 20:44:57 | |
One example is https://github.com/justinas/nixos-ha-kubernetes which uses terraform's state from a Colmena configuration | 20:45:25 | |
I don't expect provisioning | 20:46:44 | |
sounds very much like scope creep | 20:47:21 |