NixOS on Free Oracle Cloud Arm A1

Development
22

kexec’ing takes a long time, and connections can timeout during that time. Try waiting longer, it’s probably actually running kexec still while the connexion drop. In my case I had to wait a bit longer than 5 minutes, but I guess if your instance is slower for whatever reason, it could easily take double or triple that.

Rebooting during kexec will always bring back the original Ubuntu, since kexec is an in-memory only thing. It’s only when you start doing destructive operations that this may not be true, i.e. in my blog post, when you start deleting the original partitions.

2 Likes
23

In Nixos on an Oracle Free Tier Ampere machine :: korfuri's website

At the steps to format and mount

# mkdir -p /mnt/boot
# mount /dev/sda2 /mnt/
# mount /dev/sda1 /mnt/boot/

should be

# mount /dev/sda2 /mnt/
# mkdir -p /mnt/boot/ # (after mounting /mnt/ it's empty)
# mount /dev/sda1 /mnt/boot/
4 Likes
24

I was interested in deploying NixOS on my free Oracle Cloud VM. I took a different approach than a lot of the people in here.

The main differences are:

  • Instead of building a custom NixOS installation image and using kexec, I used off-the-shelf netboot.
  • Instead of interactively repartitioning the drive, I used a disko with a general purpose disk config file.

I found that these changes made the installation process simpler, though it was a lot of trial and error getting it working.

Here’s my walkthrough of what I did:

Suggestions are welcome!

7 Likes