As for previous version upgrades, I just used zypper dup to upgrade from LEAP 42.2 to 42.3. I do this simply by first editing the four main Suse repository addresses from 42.2 to 42.3, and disabling all other repositories.
The Suse update always seems to work perfectly for me on my machine, but each time I do this I inevitably run into difficulties at restart with the Nvidia drivers, resulting in “can’t start plasma” type messages. This time I managed to resolve this by restarting with the recovery mode for the previous kernel version, then using YAST to remove all installed Nvidia files and letting the system use its default video mode. I am then able to do a clean Nvidia graphics driver install.
My question though is, what should I do to get this right first time? Should I leave the Nvidia repos active when I do the zypper dup? Would this work? I could just remove the Nvidia drivers before doing the upgrade but this doesn’t seem quite right to me - there must be a more elegant method?
What has worked for my server/workstation is to indeed leave the nvidia repo active ( changed openSUSE version no in the repo URL ). This from 42.1 => 42.2 and 42.2 => 42.3
I’d like to note that exactly this did cause problems in the past though.
E.g. when upgrading from 13.1 to 13.2, I experienced this myself that the driver just didn’t work after the upgrade…
If you plan to update your distribution using zypper dup you should uninstall the old driver version first to avoid problems.
Of course the changes in 42.x were not that big, but an upgrade to 15 will be a completely different story again.
So it’s probably still recommendable to completely uninstall the nvidia packages before the upgrade.
Although it should work to install the new ones again before you actually reboot into the upgraded system.
At least be aware that there might be problems, and as first thing try to uninstall the nvidia driver completely and install it new in case you do have problems after the upgrade. (but then, the problems in 13.1->13.2 were caused by changed installation paths and files of the old version that somehow were left over and didn’t get removed when running “zypper dup” with the nvidia repo active).
Thanks Wolfi for the extra info. If I read this all correctly, the best bet looks like prior removal of the drivers, restart without them to get a working system, and then reinstall. Cumbersome but easiest (I remember you having to bail be out on this on a previous occasion :)}.
But note that you don’t have to “restart without them to get a working system”.
Installing them after the “zypper dup” before rebooting should just be fine I think.
Having the repo active while doing the upgrade will not even guarantee that the driver will be reinstalled (the package version/revision may be the same for both distribution versions), and if that isn’t done the driver likely wouldn’t work after the upgrade anyway (because of a new and ABI-incompatible kernel).
On further thoughts, that should not be a real problem though.
The nvidia kmp (Kernel Module Package), which is the relevant one here, does have the kernel version in the package version.
So the package for a newer openSUSE version should have a higher version.
Doesn’t guarantee that the driver itself gets updated, but as long the driver version is the same that shouldn’t be a problem either.
And if the driver version changes, all packages will get updated anyway.
So, likely it will work out, unless there will again be changes to the install locations (like in 13.1->13.2) .
TBH, I consider such a change yet again very unlikely in the foreseeable future.
Still, something one should be aware about, I suppose…
I gave up on all this some time ago. Everytime I upgraded while using the Nvidia binary drivers I got stuck in upgrade hell. Now I juse use the nouveau (nv) driver or buy Intel chips with the built in graphics.
Ken