I think the new kernel uses memory differently compared to the kernels in F14.
One system I just upgraded consistently crashed with checksum failures on the RPM packages during the actual install phase. In the end I found my F14 CD and ran memtest86. Massive (really massive) memory failures on a rather new PC that had passed memtest86 just a few months ago when it was built. It had been used with F14 without any big problems until now. Mostly running armacycles and supertuxcart (it's a living room PC for the kids).
Since it had crashed halfway through upgrading X wouldn't start. I logged in through ssh with X tunnelling and re-ran preupgrade after removing the offending memory module. I then had to run package-cleanup --cleandupes
I would recommend running memtest86 prior to this upgrade just to be safe. You may have defective memory you didn't know about. if you don't have a cd handy, it should be as simple as installing memtest86+ and then run memtest-setup to add it to the grub menu.
On Fri, 2011-05-27 at 11:48 +0200, birger wrote:
Massive (really massive) memory failures on a rather new PC that had passed memtest86 just a few months ago when it was built.
Were proper anti-static precautions taken when it was built?
The usual ramification of not taking them is that the components are damaged, but still work at the time. Some time later, they fail. Long time after you'd associate that failure with handling the components.
The time interval would depend on the degree of static damage. When we were studying digital electronics at college, back when I was young and dinosaurs roamed the Earth, we'd be talking about knocking years off the expected life of the equipment. i.e. When something died years later, when it should have lasted twenty years, or so, static shock would have been the diagnosed cause.
More extreme shock would have a much shorter time period, of course.