A HD decided to die on my server. Fortunately, all the important data is backed up (and I’ll describe how I do that one of these days), but the disk is still acting flaky (not sure if it was disk corruption or an impending disk error… all the self-tests check out, but I dunno….).

The old system was a Debian woody + unstable updates (3.0 + random updates)… I decided to re-install with Debian Sarge (3.1). Now, Debian, while not flashy on the UI, has always been extremely stable and relatively bug-free.

I downloaded the netinst and went with that, let the thing run overnight, and got things installed pretty easily. Then came time to start making the server like it was.

Already, I’ve found two glaring show-stopper Debian bugs… come on guys, what are you thinking?

There’s Bug #310887, Does not mount non-root partitions. Essentially, some lunkhead decided to fsck all partitions before loading any drivers, like SCSI or USB. So if you have a SCSI disk, which my previous server image was on, you’re out of luck. (note: the previous server image was on a ~10 year old 4G SCSI running RedHat 7.2 that has shown some signs of failure… but at least it’s still kicking as the impromptu backup). Haven’t figured out a work-around for this yet, but man is it a pain.

Then there’s Bug #271363 - kdm frozen at first boot. Debian uses gdm by default for gnome, which is fine. I prefer KDE and use kdm. So, I installed kdm, no problem, and switched it. But this dies a horrid death. Running things manually, I find errors about the ELF version of libkdeui.so.4… haven’t tracked down the culprit here yet either.

These are basic… sigh. But I guess people tend not to have SCSI anymore, or don’t use KDE. Um…. sure.

I’m re-installing my picture archive from backup, and will then update the new image to SID (Debian unstable… as Debian testing was the same as sarge). Or maybe I should just go back to Fedora… they seem to have gotten better about solving the rpm hell problem, so might be worth giving them a try again.

Oh, and for those paying attention — yes, these are all Linux servers, and yes, I do work at Microsoft. Why am I not considering Windows? Lots of reasons. Frankly, I don’t feel that Windows is up to the task of providing the roles I use the server for. In particular, DNS, mail serving, and file serving. Second, my main data is held on two 75G IBM Deskstar disks (still functioning after 5 years… go figure!) in RAID configuration. I’m not in the mood to consider how to migrate this over to Windows. Finally, Linux does great at headless operation, and mostly that’s what the servers are. I can ssh in, and check things out, from whereever I am. The Windows solution is this hokey Remote Terminal, where you suck a lot of bandwidth to get a remote GUI. Sometimes, I don’t have a lot of bandwidth, and all I need to do is get a command prompt and go. So, yeah, I still run some Linux boxes.