On Fri, 17 Oct 2025, Martin Husemann wrote:
On Thu, Oct 16, 2025 at 10:35:00PM +0100, Liam Proven wrote:My word. So this is in fact the _recommended_ way? And it keeps any existing user accounts etc?Yes, it is supposed to keep everything you modified that is not a base system binary - or give you choice to keep your modification (mostly for files in /etc, and it won't delete added user accounts).Do you think this would rescue my installation where I can no longer log in?Possible - depends on what corruption exactly happend. It will not make it worse :-)
Indeed. The upgrade process is not complex particularly within a major release
1) new kernel 2) reboot3) extract all sets except for etc and the kernel sets (and xetc if using X)
4) reboot 5) postinstall -s <path to etc set> fix 6) optionlly do something about etc files - I never do You can skip 2 if not doing a major upgrade. A simple (untested) manual upgrade script would be: dir=<path to sets> cd / for set in $dir/*.tar.xz do case "$set" in */etc.*|*/xetc.*) ;; */kern-GENERIC.*) tar -xpzf $set ;; */kern-*) ;; *) tar -xpzf $set ;; esac done postinstall -s $dir/etc.tar.xz -s $dir/xetc.tar.xz fixThe point here is that something went wrong with your upgrade and that's what we need to urgently fix for you. When we know what, we might be able to work out why.
So, if you cannot login as root let's try:1) boot single-user 2) enter vt100 if prompted for a terminal type
3) mount -a 4) grep ^root /etc/master.passwd If that looks healthy: cp /etc/master.passwd /etc/ptmp pwd_mkdb -p /etc/ptmp Then reboot and try again. -- Stephen