!UKDpaKNNsBpOPfLWfX:zhaofeng.li

Colmena

332 Members
A simple, stateless NixOS deployment tool - https://github.com/zhaofengli/colmena116 Servers

Load older messages


SenderMessageTime
25 Mar 2022
@github:maunium.net@github:maunium.net [zhaofengli/colmena] zhaofengli unpinned issue #50: Nixpkgs fails to build on unstable 02:22:59
30 Mar 2022
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao opened issue #68: Boostrapping from ISO
  • Suppose I have a headless machine with link-local connectivity
  • That machine has a specially prepared live iso with a well-known (static & link-local) ip address
  • The mount point of the store is to be construed at /mnt/nix/...
  • Let's call the still anonymous host (no own IP, no crypto identiy) larva
  • Now we want to: colmena apply --on larva --over [fe80::47]:2222 --store /mnt/nix

Can you give me a few pointers where to implement this?

05:11:30
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao edited issue #68: Boostrapping from ISO 05:13:06
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao edited issue #68: Boostrapping from ISO 05:13:48
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao opened issue #69: Deploying home-manager profiles

What would be the conceptual steps for feature parity with deploy-rs and enable deployment of home-manager profiles as a "grand unified deployment tool"?

05:15:07
@aanderse:nixos.devaanderse hoping someone can help me dig into this... i'm trying to deploy to a server i deployed to using colmena last week
i tried deploying to the same server today and colmena just hangs
is there a higher debug level or some flags i can pass?
09:34:22
@aanderse:nixos.devaanderseScreenshot_20220330_053229.png
Download Screenshot_20220330_053229.png
09:34:34
@schnecfk:ruhr-uni-bochum.deCRTifiedGood question, but it recently started to hang at that point for me, too 🤔 Thought that it was caused by the (at that point) unstable network connection16:37:29
@zhaofeng:zhaofeng.liZhaofeng Li After it says "Starting..." it will invoke nix-copy-closure 22:29:51
@zhaofeng:zhaofeng.liZhaofeng LiMaybe there are some environment variables that affect its logging22:30:16
2 Apr 2022
@github:maunium.net@github:maunium.net [zhaofengli/colmena] zhaofengli closed issue #62: nix-eval-jobs patch causes build failure with nixos-21.11 nixpkgs 03:03:42
3 Apr 2022
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao opened issue #70: `meta.nixpkgs`

I think this interface is not necessary and using the equivalent upstream interfaces would lead to less confusion.

  • In a flakes world, the way to determine meta.nixpkgs would be as follows: inputs.colmena.inputs.nixpkgs.follows = "";
00:56:05
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao edited issue #70: `meta.nixpkgs` 00:56:31
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao edited issue #70: `meta.nixpkgs` 00:56:45
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao closed issue #69: Deploying home-manager profiles 00:57:37
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao opened issue #71: meta.nodeNixpkgs

It is really a nightmare in terms of composability to have this attribute exposed via meta. I wonder if it would be possible instead to expose it via the normal (composable) module system and shallow evaluate for peek. something like this:

let
  colmenaNixpkgs = {
    options.colmenaNixpkgs = {};
  };
in (nixpkgs.lib.evalModules { modules = [colmenaNixpkgs]; check = false; }).colmenaNixpkgs
05:26:00
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao opened issue #72: Evaluation

I saw your efforts to speed up evaluation. This might be a project that you might keep an eye on: https://github.com/kamadorueda/santiago

Not immediately useful.

22:22:54
4 Apr 2022
@github:maunium.net@github:maunium.net [zhaofengli/colmena] Repo forked into blaggacao/colmena 00:07:31
@github:maunium.net@github:maunium.net [zhaofengli/colmena] blaggacao opened pull request #73: imp: soc of eval, modules & options
  • the reason for this change is to have more transparent separation of concern between effectuations of the module system and pre-module system effectuations

  • with improved flakes support down the line, pre-module system effectuations will get more complex

  • if the code was clean before, this does not introduce a change in behaviour. Unclean prior code, could now be surfaced by this.

00:07:43
@github:maunium.net@github:maunium.netRedacted or Malformed Event00:08:55
@github:maunium.net@github:maunium.netRedacted or Malformed Event00:10:47
@github:maunium.net@github:maunium.netRedacted or Malformed Event00:25:37
@github:maunium.net@github:maunium.netRedacted or Malformed Event00:27:46
@github:maunium.net@github:maunium.netRedacted or Malformed Event02:12:37
@github:maunium.net@github:maunium.netRedacted or Malformed Event02:12:39
@github:maunium.net@github:maunium.netRedacted or Malformed Event02:14:06
@github:maunium.net@github:maunium.netRedacted or Malformed Event02:14:41
@github:maunium.net@github:maunium.netRedacted or Malformed Event02:21:21
@github:maunium.net@github:maunium.net left the room.02:28:47
5 Apr 2022
@jhillyerd:matrix.orgjhillyerd joined the room.20:42:35

Show newer messages


Back to Room ListRoom Version: 6