On Sun, 29 Jan 2023 02:40:58 +0000 (UTC) George R Goffe via test test@lists.fedoraproject.org wrote:
Running dnf upgrade during periods of system instability seems to leave the system in an unknown state.
How does one recover from this situation?
I assume you are on rawhide, since you are asking on the test list. Rawhide is a lot more stable than it used to be, but there is still a risk when recovering from these kinds of errors. That is part of running rawhide. Have you backed up any unique information, so that if you have to do a reinstall, you won't lose anything?
Generic response. First thing, just try rerunning the update command, dnf update If that doesn't work, try running dnf distro-sync That should examine the state of the system, and correct any errors. It won't correct for hardware errors, or power failure in the middle of an update. If it still isn't working, try dnf --allowerasing distro-sync If it still isn't working, try rpm --rebuilddb and then dnf --allowerasing distro-sync
Or, before you try anything, respond with more information, so the response is less generic and more geared to your situation. What were you doing? What kind of system instability? What happened, the details? Were there error messages, what were they? You can log error messages to a file called error.txt by adding 2> error.txt to the dnf command.
It might be that you have to manually remove some packages, or get packages from koji in order to manually install them. Depending on the error, it might only require excluding packages with conflicts from update, allowing the rest of the package updates to proceed. This isn't uncommon after the mass rebuild of all packages, since some fail to build, and cause dependency errors.