I'm trying to bring up Wolverine on a machine with all of its disk behind a 3ware controller in one 538GB RAID5 array. Bugs 30731 and 28029 are related, although this machine is not running SMP. The problem is that Wolverine does not work on a RAID5 array (it works on a striped array just fine). I get large numbers of card aborts and the install runs for hours without making progress. As I would expect from bug 28029, with the boot disk and stageN image files from Rawhide[1], the install progresses fine until post-install configuration, where the installer segfaults (no Python backtrace available). I do not know if this is due to the 3ware controller or if it is simply a problem with the installer. I can boot into rescue mode using the same disk and stageN image files and see the installation. Unfortunately anaconda crashed before setting up lilo or /etc/fstab so I'll have to go manually from here. Is there currently any way to actually install any version of Red Hat with all disk on a 3ware controller? 1) I have bootnet.img, netstg1.img, netstg2.img and stage2.img from March 19. I am installing over HTTP. I've also copied the kernel RPMS from Rawhide so if I could boot the system I should be able to run.
Matt or Alan, any ideas?
We had some fixes from 3ware. I dont know which QA tree they went back out in
We believe this is addressed in the current internal tree.
Just to make a permanent record: Seawolf installs fine.