I’ve been working with computers practically all my life and thought I’d seen it all, but this one really surprised me!

At $WORK we have a number of IBM BladeCenters hosting our VMWare environment. Each blade chassis has 14 IBM HS21 bladeservers in it, and the way the environment is designed to scale everything has to be identical. Well, my department needed to take a blade chassis out of this identical environment and repurpose it for another project.

The plan was to install Red Hat Enterprise 4 on these servers using a PXE kickstart process. On other blades and equipment this would go off without a hitch. For some reason, though, these particular blades caused a kernel panic right when the Linux kernel would load. Turning off acpi didn’t seem to help. That Red Hat 5 would load on these same systems made us even more puzzled.

Finally, one of the night shift UNIX gurus had the solution, which was provided years ago by our IBM vendor. Their solution? To swap the CPUs on each motherboard! Our team dutifully did this and somehow each blade successfully booted Linux.

I about fell out of my chair when I heard that this worked. It makes no sense whatsoever but somehow it worked.

Just when I thought I’d seen everything . . .