Two show-stoppers

I’m running the nightlies, because my setup requires ZFS2.4.
It’s not production yet, so hold your horses with the usual “nightlies are buggy” comments, I know.

That said, I need to know if these are known bugs, or issues with my setup or things that have known fixes:

Promblem 1:
After each reboot, none of the datasets are mounted. So after each reboot I need to drop to the shell and issue a “zfs mount -a”, at which point all the datasets are mounted. There are no errors of any type in the pool. So it’s not some zfs pool error that prevents mounting.
The upshot of this, is that of course no apps (most problematically not Tailscale) and containers can auto-start, because the datasets they rely on aren’t there.

Problem 2:
After each boot the Apps system magically forgets its configuration. So each and every single time, i must be unset the pool, and then set it again, at which point all the old apps are recognized and (where applicable) deployed as they should have been. I know there’s a certain overlap with Problem 1, but in a previous test installation, I had the same problem, at a time when I had apps installed in a way that they wouldn’t rely on my data sets, and it was the same issue.

The upshot is, I can’t access the system except from the LAN if anything causes a reboot or if I want to update the system while not locally present, because without zfs datasets mounted, and the app system remembering its configuration, Tailscale won’t launch, meaning I’m locked out from anywhere but when on the LAN. This makes things rather cumbersome, even for a test setup.
Particularly Problem 1 is bugging me, because without that, I could if need be install tailscale in a container rather than as an app.

This has not been seen in house, perhaps you could raise a bug ticket so we can understand what is different with your environment

On the Jira system?

Yes