I went to switch channels and upgrade. The upgrade broke due to the version of systemd in unstable being at 256.8 and the version in stable being at 256.10.
Can anyone tell me who to address so this can be remedied? At present, it is likely not possible for anyone doing a switch to unstable, and upgrading for the first time on a clean install to be able to do so due to this.
I asked on Matrix and systemd
257.1 is on staging
, and staging
is merged on master
branch twice per month.
Yes, but I would prefer to use the unstable channel instead of staging or master.
I never recommended switching to staging
, I’m just saying that this state discrepancy is because of staging
branch to master
cycle.
Yeah, do not use staging
.
It’s pending Systemd 257 by jmbaur · Pull Request #356818 · NixOS/nixpkgs · GitHub to make its way into master
(which will be merged from staging-next
to master
within the next week presumably). This happens via staging-next 2025-01-06 by vcunat · Pull Request #371501 · NixOS/nixpkgs · GitHub, and manual testing is generally done on the staging-next
branch to verify that important packages are working as designed prior to the merge.
Status is available on the PR tracker https://nixpkgs-tracker.ocfox.me/?pr=356818 and you can switch over once nixos-unstable
shows a checkmark.
Understood @drupol and @waffle8946 and thank you for clarifying. But in order to avoid the issue, why don’t they just add 256.10 in the meantime so people can at least upgrade.
Because the PR still has to go through staging
→ staging-next
→ master
(because systemd
upgrades cause mass rebuilds) and that wouldn’t speed anything up here. And the intention is to not break stuff willy-nilly.
Two updates were prepared: One to 256.10, the other to 257.1.
Both need to pass through staging: One through staging-next, the other through staging-next-24.11.
Based on the order in which staging-next and staging-next-24.11 were processed we shipped 256.10 to nixos-24.11 first, and are shipping 257.1 to nixos-unstable later this week.
Broke how?
If you’re doing a major upgrade, I’d personally recommend nixos-rebuild boot
and reboot, not nixos-rebuild switch
.
I used nixos-rebuild boot. It wouldn’t complete due to the package in unstable being lower than the one in stable
Oh, the bootloader stuff. To me that sounds like a problem in itself, as we surely want to allow downgrading – say from nixos-unstable to a (maintained) stable.
I am not sure if you are missing the point from my end, so please pardon my expressing it again.
What I am expressing is that I am having it stop with an error when I try to switch from stable to unstable because the stable branch has 256.10 and the unstable branch has 256.8. So as it is trying to switch from 256.10 to 256.8, it brings a message up that it is already at the latest version and then exits the upgrade with an error message expressing so. Then when you reboot, it is still on the initial version.
As I mentioned here - New install: unable to switch to unstable because systemd-boot version - #2 by waffle8946
You can delete the files mentioned in the error and reinstall the bootloader, if you simply must upgrade now.
My system is showing nothing in the boot folder and will not let me terminal into it.
Well it’s not necessarily the same directory, your error will be different if your ESP is mounted elsewhere. But you do need root to perform this operation most likely.
No problem. I believe I understood before.
I just pointed out that while your problem is transient due to race between two staging* updates, rollbacks will have the same issue most of the time and waiting won’t help. So it would be nice to have some solution.