You call it moving the goal posts -- I call it dynamic e-business
People in the field are telling us that they are getting a bunch of 'predictive memory failures', most likely errors that are trapped by ECC or other correction mechanisms. IBM is replacing the failed DIMMs without explanation, and has offered BIOS upgrades and new DIMMs to cure what ails the P Series boxes.
Now I wonder why you would need two firmware updates this quarter to fix a DIMM failure, if it is fixed? Wouldn't a bad DIMM just need a good one to fix it? This one reeks of something, most likely fish.
To make matters more interesting, all of the replacements for the failed 4GB DIMMs are being shipped out of Rochester or Poughkeepsie, not out of the local distribution centers. There are also no parts in the channel. IBM will also not release a list of the failed parts.
So, what do we have? Random failures, patches where there should not be any, recalled replacements, and parts from the mothership directly. Luckily, IBM isn't telling its customers of these high end and shatteringly expensive boxes WTF is going on. How nice of them. If you have had this happen to you, let us know please. µ