I had a really bad hardware day today. I'm not really sure what happened, but one of my systems had a harddrive that started having read errors. It's a pretty old PII with a 2.5GB Maxtor HD (and a 340 MB Western Digital that has been rock solid for over a decade), and 128 MB of ram. The BIOS is kind of flaky, and it can be hard to be sure whether I'm having hardware, or configuration problems with it. reiserfsck was telling me I had bad blocks, which may, or may not actually be true. The errors were happening near the end of the check process, which may mean the physical peremeter of the disk. I tried rebuilding the tree, but that failed, and I got an error (IIRC, from reiserfsck) after that saying that there was a problem reading block 0. Well, I had already decided to cannibalize another old system and add a harddrive to this one. I decided to slick this harddrive, and reinstall the OS, using the problematic HD as the root. (I was betting the read errors weren't really from hardware, but from the BIOS not being configured correctly.) It seemed to take the installation OK, but I couldn't get YaST to install the bootloader, I couldn't even get it to install it to a floppy. What happened was that it would run mkinitrd, and then say it was copying the configuration files. I could hear it grinding away for a while, so I suspect it was doing something like that. But then it would just sit there waiting to start actually writing the bootloader to disk. (I *believe* that was the third and final stage). I looked at top from tty2, and saw a few processes going which seemed related to this. One of them was ag_bootloader. After about a minute, it would jump to over 90% CPU utilization, and stay like that for over 20 minutes. Note that I was able to boot the system if I first started the installation process from CD. If I tried to boot directly from the HD, it just printed GRUB and sat there. I finally rearranged the harddrives so the new one is the root, and the problematic one holds /opt. (This box is mostly used for stuff like SVN, so a broken KDE won't kill me.) The system installed OK with the new arrangement. But I'm still curious as to what was wrong. I was able to dd if=/dev/hda1 of=/boot/current_mbr bs=256 count=1, so it seems the MBR was accessible. Does anybody know what ag_bootloader is, and why it would behave as I described? -- Regards, Steven