Reinstall Linux?

[Michael Kennedy]

Well, with me if it's not one thing, it's another.  I updating gcc on my Mandrake box when I got a segmentation fault error.  My heart sunk a bit because I knew that was most likely a hardware problem.  I had an odd configuration on the box since I had 2 sticks of EDO and 2 sticks of SDRAM in there for giggles and all ran well up to that point, so I decided to go ahead and remove the EDO to see if that was giving me the problem.  Well, after that the whole system completely failed to show any signs of life when I flipped the switch back on.  I chalked it up to a dead power supply (my new theory on the signal 11 errors) and began transplaning everything into a bigger case (I decided to add a few HDDs and a Promise card in the process).  To my horror I then saw what I will officially deem the culprit in my error laiden Linux box- a horribly leaking capacitor that barely seems to even be attached to the board.

Soooo, that motherboard is now gone.  I do have a 700 K7 that will once again become my Linux server.  Now comes my long awaited question- do I have to reinstall Linux to get the server back?  The old CPU was a 450mhz K6-2 and on a completely different motherboard (of course) and I bet most items are compiled to run on that.  At the very least I bet a kernel recompiling is necessary if only to take advantage of the new features of the mobo and CPU.

I'm done with this #OOPS# for the night, so any suggestions anyone can throw my way be morning would be great.  Thanks.

[William Grim]

You don't have to format your hard drive to get it working again, assuming you are careful during the transition.  I'm something of an experimenter when it comes to software and development stuff, so here's what I'd do.

I would take your new computer configuration and put it together.  Then, I'd take the boot CD to your distribution of GNU/Linux (yeah, too many ppl get pissy about the proper use of the OS name; so, I'll just start saying GNU/Linux on here if I remember) and boot up the image.  However, if you have a default kernel already installed on your system called "safe-mode" or whatever, boot that linux image instead.

Once you have the system up, if you're doing it from CD, mount your hard drive and chroot yourself into it.  Then source /etc/profile and any ~/.profile or ~/.bashrc's you might have (or any other relevant file) in order to setup your PATHs, etc, correctly.

Now you have a working system running, almost like if you booted it normally.  As long as your gcc is installed correctly, along with all associated libraries, you can go to where you keep your kernel source code (and hopefully your configuration file as well) and "make menuconfig'.  Then, you can change the kernel to your CPU type and make sure the rest of the kernel is correct, save, and compile.  Then follow the normal steps to installing the kernel, modulees, etc, and rebooting.

Now you have a system that's ready to rock and roll, assuming you don't have to replace too many specialized packages that were built for your old CPU.