Opensuse-migration-tool fails

[pinguin74]

Ok, tried now to upgrade 15.6 to 16.0 using the migration tool.

It stops at package no. 1484 out of 10186:

Installation of filesystem-84.87-160000.2.2.x86_64 failed:
Error: Subprocess failed. Error: RPM failed: command exited with status 1.

Looking at journalctl doesn’t give more clue:

scriptlet %prein(filesystem-84.87-160000.2.2.x86_64) failure: 2

Did this package have a problem with existing file system topology?

Since this occured in the middle of installation I might have a broken system now and dont dare to reboot…

Thanks for any hint

Btw, before using migration tool I wiped all repos and installed the minimum repo set using zypper in openSUSE-repos-Leap-20250714.a450212-lp16.0.3.1.x86_64, so no repo weiredness.

Thanks

[pinguin74]

I now erased the symlink /usr/lib/modules and it made the installation to continue, though that error message appeared again. Let’s see, if the upgrade succeeds…

[pinguin74]

Nope, stuck again, now at
“warning: Converting database from bdb_ro to ndb backend”

The thought of doing a clean installation from zero comes closer and closer.

[john_hudson]

I had the same problem and I solved it by saving the output of opensuse-migration-tool --dry-run to a file and then entering each of the commands in the file with my own ‘adjustments’ to the commands - in particular I changed the dup command to install files one at a time and not try to download them all before installing them. This solved the problem for me.

[pinguin74]

I now found the underlying problem, Im on Lp 15.6, but the RPM data base is much, much older from previous installations, that makes the process hang at some point. Thus, this issue affects mostly, or only machines that were setup years ago and carried their RPM data base from distro upgrade to distro upgrade.

The whole issue with a workaround is described there:

https://bugzilla.opensuse.org/show_bug.cgi?id=1242188#c10

The workaround is described in the first posting and consists of converting the rpm database in a 2nd shell, when the upgrade hangs and continue the upgrade after data base conversion, which might take hours! In my case almost 3 hours.

[C7NhtpnK]

@pinguin74

I now found the underlying problem, Im on Lp 15.6, but the RPM data base is much, much older from previous installations, that makes the process hang at some point. Thus, this issue affects mostly, or only machines that were setup years ago and carried their RPM data base from distro upgrade to distro upgrade.

The whole issue with a workaround is described there:

https://bugzilla.opensuse.org/show_bug.cgi?id=1242188#c10

Thank you for this hint! Appreciated!

How do I find out, what I have currently now? I.e. if there is need for converting? TIA!

[knurpht]

I sort of ran into this with a customer laptop, that has runs openSUSE from an SSD from a previous laptop, likely from ~2010. A simple sudo rpm --rebuilddb fixed it. No trickery needed.

[C7NhtpnK]

@knurpht

A simple sudo rpm --rebuilddb fixed it. No trickery needed.

Are you sure? Have you actually checked it?

rpm --showrc | grep db_backend or rpm --eval '%{_db_backend}'.

---

@pinguin74

The whole issue with a workaround is described there:

https://bugzilla.opensuse.org/show_bug.cgi?id=1242188#c10

The workaround is described in the first posting and consists of converting the rpm database

What is sadly missing there:

/usr/lib/rpm/macros:
%_db_backend bdb →→→ %_db_backend ndb

---

I have discovered this because it was the case on my system before. I have Leap 15.6 (close to upgrade to 16.0, now), having this system since 2020 initially set up then and always updated inline.

[knurpht]

Can’t check, laptop is not here, something must have changed it at some time, but not me. I could imagine some posttrans script failed, but no way to find out anymore.