Looking at the logs it seems that anaconda somehow ends up consuming all the memory on the system (2Gb) and is killed. This happens a couple of minutes after the message "Moving to find rootparts" in the anacodna log. Before that there are only some seemingly harmless warnings about using a deprecated SCSI ioctl.
I've tried using both graphical and text mode installs with no difference. I've done a full fsck on the partitions which all passed OK. The system is back running FC2 with no ill effects and no signs of any problems, and now I'm stumped as to what to do next. The really strange thing is that our database server, on pretty much identical hardware (Dell PowerEdge 2650) upgraded without any problems yesterday.
The only clue I've managed to turn up as to what may be causing this, is that during the fsck the system tried to scan for lvm volumes using vgscan and this caused a segmentation fault (which is reproducible once the system is running). There are no lvm volumes on this machine, but there are on our database server - so maybe this is something?
I tried to remove lvm2, but there are too many dependencies so I reinstalled it, but it still segfaults.
Can anyone suggest how I can proceed with this??
Cheers
Simon.
