Resolved by upstream commit: https://github.com/upgrades-migrations/preupgrade-assistant/commit/a01f099a2b4237e267d902c5e7e3d437e8710ca3
I cannot reproduce this.
This was originally found in 2.2.1-1.el5. On el6 channel, the last version is 2.1.10-6.el6, so my guess is that the problematic code was never released.
Michal, can you please confirm if that is the case? If so, I would propose dropping this from erratum and closing it (there is nothing to fix from RHEL6 customer POV). Otherwise, please provide hint how to reproduce this on RHEL6.
Alois, to reproduce this in 2.1.10-6.el6, make sure you have "autopart" in /root/anaconda-ks.cfg - it serves as a basis for the generated kickstart.
When generating kickstart the "part" commands were always added no matter whether there was "autopart" already or not.
Sorry, this does not work: 2.1.10 still passes, no errors seen.
(It did work for bug 1414341, where I have asked very similar question; sorry if I made it confusing.)
Ok, it's possible that this bug, describing different results when executing the same command twice, is related to a development version of the Preupgrade Assistant. I have no other ideas on how to reproduce it. I'll close and drop it from the Erratum.