Strange bootloader issue with Debian Lenny.
Hi, all! I've just began experiencing a really weird issue with my Debian server and hope someone can help me out.
I'm using an IBM NetVista workstation with four IDE hard drives (no CD-ROM) as a general Debian multi-purpose server (Files, DHCP, DNS, PXE preboot environment). All of the hard drives on the system, including the primary one, were used as bootable system disks on other systems but are now strictly for storage. Airflow to the drives seems pretty good, as temperatures didn't seem to be at dangerous levels last time I checked (between 29 and 41 deg C).
I came home one day to find that my server's network connection went down, so I restarted it thinking nothing of it. Unfortunately, it stopped booting when it came back up. After hooking up a monitor to it, I noticed that it POSTs successfully, but will boot to a blinking cursor and nothing else. However, it DOES boot if I disconnect the second IDE channel from the motherboard and the slave drive on the primary channel and reconnect them all again. (It took about 3.5 hours, a replaced IDE cable , lots of mix and matching and buckets of anger and frustration to figure this out.)
I tried a couple of things, all to no avail. I looked at to see if it was targeting anything other than (hd0,0), but it wasn't. I also removed quiet boot option while I was there. and all identified (hd0,0) as the system volume (master device on IDE channel 0). I figured that the other boot drives might still be flagged as bootable (I got NTLDR missing from a disk that had Windows Server on it before), so I disabled all drives with that flag enabled.
I'm pretty set that GRUB is setting another drive as (hd0,0) somewhere during boot or shutdown, but I have no evidence suggesting that this is the case. Has anyone seen this before?