The installer hangs during an upgrade from RH6.1 to RH7.1 when the fstab includes /dev/md0. Workaround, comment out the /dev/md0 line before upgrade.
Created attachment 16986 [details] Dump from anaconda saved during upgrade
Can you attach your /etc/fstab and /etc/raidtab files?
Created attachment 16989 [details] raidtab and fstab from upgraded box
The raidtab and fstab are not exactly the same as when I did the upgrade, but very nearly so. The the workaround actually involved commenting out the /dev/md0 line in the fstab, temporarily removing the raidtab, and disconnecting the RAID disks from their respective controllers. Since I am booting from a SCSI disk, this worked for me.
Where did it hang? Raid upgrades work on our machines here.
It appears to hang during the mount of the file systems. I cannot remember exactly when the error message came, but it was after I had chosen the upgrade option and before any files were actually copied to the system disk. There was enough information in the error window that popped up to give me the idea of commenting out the /dev/md0 line in fstab -- something about failing temporary mount of multi-disk device. The message also asked me to save the anaconda dump and submit the bug report. I assumed it would have all the information about what was going on at the time. Could the problem be related to the fact that this is an IDE RAID with a SCSI boot disk? This seems to me to be a fairly unusual setup.
Yes, it could have something to do with IDE RAID with a SCSI boot disk. Does the behavior change if you boot the installer by saying 'linux ide=nodma'?
That I don't know. Once I got the upgrade to happen I was happy. If you want to experiment with this, it's a Promise Ultra 66 PCI card in addition to the Tyan S2390A onboard IDE contollers with 4, 75GB IBM DeskStar 75 GXP disks connected one per IDE channel (80 wire UltraATA cables). The SCSI is a BusLogic BT-946C controller with one, 18GB IBM DNES-318350W, and an external Toshiba cdrom.
Ok. I'm glad you got the upgrade to work. I'm pretty sure that this was caused by the some devices having troubles with dma transfers. We've seen a few of those. Thanks for your report.