question on a kernal "uhhuh" message

[Paul Newell]

Tim wrote:
> Paul Newell wrote regarding memtest86 results:
>   
>> I ran the test and it immediately failed. To make sure I was doing
>> things right, I tried in on another machine (which worked) and am
>> trying it on a third machine (which I will check in on tomorrow
>> morning)..
>>
>> Though I am not seeing any problems running on the machine that
>> reported the error during yum update, I am figuring "I gotta hardware
>> problem".
>>     
>
> One of my Frankenstein boxes, made from left over computer parts, had
> faulty RAM.  The system would often run fine, there'd be occasional
> faults.  I didn't know it had faulty memory at the time, it didn't have
> Linux on it, and had never been subjected to the memtest86 program.  It
> was much later on that it got tested, and it failed the test every time.
> As an experiment, I left it running as a webserver.  It continued to
> work most of the time, but would occasionally die.  It must have just
> been luck that the faulty part of the memory wasn't being used, or was
> being used by something that didn't notice the fault.
>
> Moral of the story:  A computer that apparently works fine can still
> have faults, you've just not seen them yet, or not associated some
> apparent software faults with actual hardware faults.
>
>   
Appreciate the comment, thanks. I can live with a fault if I know 
whether it is going to bite me or not later on. I've sent the memtest68 
output to manufacturer to see if I can get more info (understanding that 
their assumed stance will be "must replace" ... I'm just hoping to get a 
"what is it?").