I am interested to hear what people think about this. I am running Sarge and tonight I tried to upgrade my kernel (from a home-rolled kernel built from kernel-source-2.6.8 to kernel-image-2.6.8-3-k7). All I can say is "Wow," and I mean that in a bad way.
First, my machine was left unbootable because the install got the Grub configuration completely wrong. After booting an Ubuntu LiveCD, I was able to get at the backup of the menu.lst and manually fix the incorrect hd(?,?) and root= entries. After that, I tried to boot the new kernel. When it came time to bring up the RAID devices (there are three RAID1 arrays), the numbers had changed (e.g., md1 became md0). Of course, this made it impossible to mount some of the filesystems. Try to reboot to the old kernel. The RAID arrays maintained their incorrect numbers. Back to the LiveCD. Forcibly reassign the minor numbers. Boot the old kernel again. All manner of services segfault and have errors on startup. Boot the LiveCD again and forcibly fsck every single filesystem. Get some errors on one or two filesystems, but dozens of errors on /var. Fix all the errors and reboot. Same segfault and missing file problems on /var. Reboot LiveCD and re-fsck /var. More problems. Repeat several times. Finally, boot LiveCD and mount partition with free space, rsync /var data over to temp directory, nuke filesystem on /var and re-rsync everything back. Boot old kernel and things seem to be working for the time being. I am wondering if anyone can give me any clues as to what is going on. I have never had such an experience with Debian. I am just troubled that a simple kernel upgrade has the potential to cause so much trouble. -Roberto -- Roberto C. Sanchez http://familiasanchez.net/~roberto
signature.asc
Description: OpenPGP digital signature
