Minimal nixos-config with disko
#1
Conversation
|
Is this already deployed, or waiting to be deployed? |
Already deployed |
In that case, let's add to README the instructions for deployment. As a next step, we'll had this over to another person for doing deployments for their machine. |
qemu-guest module is common configuration for VMs (imp. for kexec) not-detected module is for non-free devices not detected by `nixos-generate-config` (might not be necessary in our case)
README.md
Outdated
| | 3. | In `disk-config.nix`, set `disko.devices.disk.main.device = "/dev/<device-name>";` | | | ||
| | 4. | | `nix run github:nix-community/nixos-anywhere -- --flake .#office root@<ip-address>` | | ||
|
|
||
| **Disclaimer**: The `<ip-address>` might change in kexec mode or post-installation, in which case you will have to stop the `Step 4` and rerun with updated `<ip-address>`. |
There was a problem hiding this comment.
Can we clarify this further?
When you say stop the Step 4, what is meant exactly? At what point (during the nix run ... process) exactly should the user stop it with Ctrl+C? Or is it that they must wait for it to error out (in which case, there is no need to stop the nix run ... process because it is already stopped), and then re-run it? If so, it would be useful to paste that error message here.
There was a problem hiding this comment.
Also how do we find out the new ip address? It is worth including that as well.
There was a problem hiding this comment.
(Disclaimer: I might be missing something, so we can also revisit these instructions after the actual install ....)
There was a problem hiding this comment.
At what point (during the nix run ... process) exactly should the user stop it with Ctrl+C?
When the ssh starts timing out. I will add that point
|
FWIW, trying this PR (with changes to ssh key/ disk path) on Digital Ocean shared droplet: 
SSH just hangs, even after restarting the droplet. 
It is possible that the bootloader configuration needs a change for Digital Ocean. This may be worth making a note of in README (we are so far assuming a particular i9 desktop configuration). |
Could be because your instance might not support EFI firmware? is it possible to run this on the droplet instance: If that's the case, we can try by removing the EFI related lines from the configuration. It should not use EFI by default: https://search.nixos.org/options?channel=23.05&show=boot.loader.grub.efiSupport&from=0&size=50&sort=relevance&type=packages&query=efiSupport |
|
Digital ocean droplet instance wouldn't boot because of the absence of qemu-guest.nix module. After using this, it does boot up but fails to automatically detect networking configuration (this could be because it is a VM instance). Solutions like nixos-infect work by generating a |
WIP blog post: juspay/flakular.in#25