I can confirm this bug. Happend on ine of my three systems. I thought it was a problem with running out of free space but after stopping the update process and rebooting the machine everything was alright. I only had to delete/reinstall telegram-desktop for some out of date dependency.

I think it is a very important issue. No response in 2 days. I will plan a switch to freebsd soon

Was the update list empty?

Also, which desktop was it, MATE or XFCE?

I ask the latter question because XFCE had problems with the last upgrade.

I'm sorry for not getting back to you sooner. I have been sick with COVID for almost two weeks.

I moved the issue to the proper issue tracker.

First of all: Get well soon!!

I have MATE. Where can I find any logs covering updates? I always forget, because it (almost) always works.

If there was an issue in will ~/update.failed.

No update.failed file.

I can only state I removed Linux support first, that one did restart the updates. But there are more places. In the end I simply looked at where it restarted the process and simply stopped it with a reboot, that helped but I somehow feel bad running an os after such problems.

I have investigated the loop issue and it must be the self-repair. If a package fails to upgrade because of a package that won't upgrade, Update Station uninstall it, reinstall it, and restart the upgrade.

I will investigate further.

In relation to issue #95 do you advise to update to 24.01.1 using pkg to be able to see which package causes the problem?

@q-pa yes

I think I should remove the self-heal code and revisit that in the future.

I am having problems with this on my laptop with the upgrade to 24.01.1.
It worked with my two desktops (after getting update-station working #143
I noticed it it restarts after a random number of updates.
I ran pkg upgrade -f and it's still random sometime less than 200 of 1400 or so but sometimes over 1000.
Each time it fails it is chown or chmod complaining about a bad file descriptor. The file in question looks like a temporary file (sensible name dot bunch of random characters).
I tried deleting sshpass, which was one of those.
Ran pkg upgrade -f a number of times till it got all the way up to ports and it froze.
Reboot and no networking.
Am now trying again from a BE and deinstalling ports and sshpass first.

It took a few tries. Still got the error but eventually it got through all the updates.
The other thing I did is I had it plugged into an ethernet cable. The other two PCs are desktop - non-wireless.

It seems pkg upgrade -f only upgrades 23.10.1
After doing that update-station still wanted to do a major version upgrade.
I tried again and then realised I should follow the instructions :-
sudo env ABI=FreeBSD:14:amd64 IGNORE_OSVERSION=yes pkg-static bootstrap -f
sudo env ABI=FreeBSD:14:amd64 pkg-static upgrade -f
I opened Firefox to get those instructions and it took a long time then the system killed it along with the running pkg.
It seems to be a memory problem but the laptop has more ram than the other two PCs! (12288MiB vs 10240MiB and 4096MiB).

My final solution was to go back to the last good BE but not login to the GUI. Instead I went to a console.
Tried as root but it didn't seem to be able to reach the ghostBSD server! (I think root uses bash shell but I use fish - don't know why that might make a difference)
Tried as myself using sudo as above and everything just worked.

It should be better with some of the latest versions of Update Station; I realize that with a forced upgrade, the selfheal would try to reinstall everything on each package fix.

To fix that once and for all, I think we need to identify all packages that depend on the kernel version and force upgrades only for those that depend on kernel version changes.

It would be easier to identify all those that don't depend on the kernel version. You could use such a list before it is complete whereas you would have to wait for the list of those that do depend to be complete.

I think I finally figured out the problem. I am working on a fix.

As requested via Telegram please find attached a video of the loop going into another round.

OK, video is too big... So I will send it in Telegram

I have made a change, but I can't test it right now. It will remove the force upgrade. Everything loops with the self-heal.

Very cool, really looking forward to it. Unfortunately, I can't test anything myself until end of next week earliest, because I am not near my machines.

Is this part of the latest updates? Because I would be able again to test it.

I have one part completed, but I need more testing.

Meanwhile, I managed to update all my three systems from 24.04.2 (pkg-base) to 24.07.3 by following these steps:

1. create backup BE
2. boot live USB system
3. set up internet connection (wire preferably)
4. sudo zpool import -> check that original zroot pool is available
5. check whether /mnt exists,otherwise sudo mkdir /mnt
6. sudo zpool import -fR /mnt zroot
7. zfs list (check)
8. zfs mount
9. ls -1la /mnt (complete system available?)
10. zpool get bootfs shows current BE
11. sudo zfs mount zroot/ROOT/...current BE...
12. don't know if that is necessary but for all datasets listed with "zfs mount" I did:
    sudo zfs unmount ...
    sudo zfs mount ...
13. make sure the whole system is mounted under /mnt
14. sudo pkg-static -r /mnt install -y -r GhostBSD-base -g "GhostBSD-*"
15. sudo zfs unmount -f zroot/ROOT/...current BE
16. sudo zpool export zroot
17. reboot and remove live system USB drive
18. maybe one needs to update the EFI files afterwards...

So, I have good news. Pkg now correctly supports upgrading kernel changes, and I can clean all the complex codes to upgrade them properly.

It looks good so far. The problem will likely be solved once and for all.

Good to hear! I only had problems when updating from 24.04 to 24.07, though. Later updates just worked.