| 31 Mar 2024 |
 @janik0:matrix.org | In reply to @technicus:matrix.org
How can I add a mechanism that will backup the directories whenever I do sudo nixos-rebuild switch? just overwrite nixos-rebuild with a wrapper script in your path or have your own deployment wrapper (can be as simple as a bash script) | 23:28:41 |
 Miles Dyson | In reply to @raitobezarius:matrix.org
too many files just make it tedious What is a method to make it less tedious? | 23:30:00 |
| Miles Dyson | In reply to @janik0:matrix.org
just overwrite nixos-rebuild with a wrapper script in your path or have your own deployment wrapper (can be as simple as a bash script) Thanks! | 23:30:15 |
 adamcstephens | Make a systemd oneshot without wantedby, and start it on each nixos-rebuild | 23:30:17 |
| adamcstephens | Then you can also schedule it with a timer :) | 23:30:41 |
| Miles Dyson | In reply to @adam:robins.wtf
Make a systemd oneshot without wantedby, and start it on each nixos-rebuild I am terrible at making systemd units. | 23:31:06 |
| @janik0:matrix.org | In reply to @technicus:matrix.org
Thanks! (most people use nixos deployment tools like https://github.com/zhaofengli/colmena but that only makes sense if you start managing more then one device) | 23:31:43 |
 raitobezarius | In reply to @technicus:matrix.org
What is a method to make it less tedious? group more stuff together? | 23:32:17 |
| @janik0:matrix.org | In reply to @technicus:matrix.org
I am terrible at making systemd units. just track your config in git and play around until you figure it out, this is one of NixOS strengths when it comes to learning stuff. | 23:32:38 |
| @janik0:matrix.org | In reply to @technicus:matrix.org
I am terrible at making systemd units. * just track your config in git and play around until you figure it out, this is one of NixOS strengths when it comes to learning stuff (because worst case roll back is really simple compared to other systems) | 23:33:06 |
| adamcstephens | This is about as simple of a unit as you can make. Possible with just `script` and maybe `path` attributes | 23:33:30 |
| Miles Dyson | In reply to @janik0:matrix.org
(most people use nixos deployment tools like https://github.com/zhaofengli/colmena but that only makes sense if you start managing more then one device) I do intend to manage multiple devices.
Are there multiple tools?
| 23:33:34 |
| @janik0:matrix.org | In reply to @technicus:matrix.org
I do intend to manage multiple devices.
Are there multiple tools?
yes there is like 20 deployment tools, everyone and their dog built one 🙈 | 23:34:02 |
| adamcstephens | Oh no should I not be building a deployment tool? | 23:34:41 |
| @janik0:matrix.org | In reply to @adam:robins.wtf
Oh no should I not be building a deployment tool? please don't, at this point nixos deployment tools are the equivalent of javascript frameworks in the nix world. | 23:35:24 |
| @janik0:matrix.org | what are you missing from other ones? | 23:35:34 |
| adamcstephens | Pull based, no building or eval on end systems | 23:36:05 |
| @janik0:matrix.org | In reply to @adam:robins.wtf
Pull based, no building or eval on end systems there was one posted on discourse recently doing exactly this iirc | 23:36:24 |
| adamcstephens | Cachix deploy can technically do this, but it’s proprietary | 23:36:32 |
 hexa | comin | 23:37:01 |
| hexa | https://github.com/nlewo/comin | 23:37:06 |
| @janik0:matrix.org | In reply to @hexa:lossy.network
comin yeah that one, I forgot the name. | 23:37:21 |
| hexa | but you'd probably need a CI and a cache? | 23:37:26 |
| adamcstephens | And it still evals, no? | 23:37:53 |
| hexa | most likely, yes | 23:38:07 |
| adamcstephens | Even if you build on CI and push to cache | 23:38:12 |
| hexa | you want something that pushes the closure to the machine and switches | 23:38:37 |
| adamcstephens | Pulls but yes | 23:38:48 |
| @janik0:matrix.org | wait how would you do the deployment part without evaling anything on the remote host 🤔 | 23:38:52 |
| hexa | hmm | 23:39:10 |
