unexpected logout

[Bob McConnell]

Dj YB wrote:
> On Friday November 13 2009 02:44:54 suvayu ali wrote:
>> 2009/11/12 Ed Greshko <Ed.Greshko at greshko.com>:
>>> Ed Greshko wrote:
>>>> Yes, it is due to the extension and the way it is treated by MIME....but
>>>> that's another story.....
>>>>
>>>> What is normally recommended is that you run memtest86+.  First you have
>>>> to install it as it isn't installed by default.  Then, AFAIK, the best
>>>> way to run it is standalone.  Meaning from boot.  As an example, I have
>>>> the following in my /boot/grub.conf file.
>>>>
>>>> title Memtest86+
>>>>     root (hd0,0)
>>>>     kernel /memtest86+-2.11 ro root=/dev/mapper/vg_f11-lv_root rhgb
>>>> quiet
>>>>
>>>> Also, I have
>>>>
>>>> timeout=8
>>>> splashimage=(hd0,0)/grub/splash.xpm.gz
>>>> hiddenmenu
>>>>
>>>> to ensure I can get to select what I want....  :-)
>>>>
>>>> Since the problem is intermittent, you should leave it running for an
>>>> extended period.
>>>>
>>>> Ed
>>> The other option I forgot to mention is to download the bootable CD
>>> image from http://www.memtest86.com/ and boot from the CD that you burn.
>> ... or just boot with a Fedora CD or DVD and select memtest in in the
>> boot options menu. :)
>>
> thanks
> but how much time should I wait?
> could there be another explanation?
> I will run this test now and get back to you after it is done
> cheers
> YB.
> 

How frequently was the error occurring? I would wait at least twice that 
long if memtest doesn't report any errors. The computer will be off line 
for that period. That's why I just replace the memory cards. I can put 
the computer is back in service. If the error returns after that, I am 
no worse off than before, but then I can decide whether to replace the 
box or take the time to do further troubleshooting.

Make sure you run the test under the same conditions as the computer 
normally operates, covers on, in the rack, etc. Memory is very sensitive 
to a number of environmental and functional variations, including 
temperature, write/read sequencing and even address sequences. Testing 
it on the bench with the covers off will change those conditions 
significantly.

Many years ago there were articles published about how to create schmoo 
test patterns to detect errors that only occurred with specific 
densities of '1' bits in sections of the die. Each test was specific to 
one particular memory chip. Now that the manufacturing processes have 
matured, I believe that class of tests is only used in production prior 
to cutting the chips the from the wafer. However, memtest probably 
includes some of those tests.

Bob McConnell
N2SPP