- Maybe the hub I was using was faulty - tried a cross-over cable instead. No better.
- Maybe a corrupted CD - burnt another - no joy.
- Maybe a file was corrupt on kickstart server - rebuilt the repos - no change.
- Maybe the year on the shelf had restored an old bad cmos setting - checking - but nothing relevant could be modified.
- I tried doing a download loop on the "kdebase" RPM (one of the ones that was failing). Every time it was downloaded it had a different check sum, and after 3 or 4 attempts the machine would report HDU errors then crash.
- I tried one of the Dells - that worked file.
- I tried using a PCMCIA card in the Acers - that worked too.
I've been working this weekend on getting the various old Laptops etc we use for the Lark-in-the-Park registration system back up and running. They've been sitting on shelves for a year, and prior to that did their two weeks in Sidcup and two weeks in Ramsgate under a combined attack from damp and dust. For what follows to make sense you need to know that, for historical reasons, we are using Fedora core 6 un-updated for these machines. Yesterday I pulled out the first of 12 Acer travel mate laptops, wiped the hard drive and started a "kick start" installation. After about 40 minutes the screen filled with garbage, and the machine died. Odd! I tried again - this time the installation worked, however when I ran the "litp-install" script which does a collection of "yum" installs and custom edits, a number of the larger RPM files failed to download properly. A second attempt threw hard disk errors, and then the system "oops"ed. "Oh well" I thought - that's one less machine this year. I then tried the next one and had the exact same problems, ditto the third and the fourth. Something odd but consistent was going on. I tried everything I could think of to solve the problem.