I once had a machine that, if you ran the memory test continuously, it would 
succeed for about 6 hours, fail once then go for hours again before falling.

Bulging our leaking capacitors are another good bet as Jay says. Those are 
prime suspects, particularly if a piece of equipment works fine in a rack for 
years then fails after you turn it off and back on.

Bests,
Paul.



Sent from my mobile device. Please excuse my brevity and any typos. 

<div>-------- Original message --------</div><div>From: Leo 
<[email protected]> </div><div>Date:13/05/2015  23:42  (GMT+00:00) 
</div><div>To: Hampshire LUG Discussion List <[email protected]> 
</div><div>Subject: Re: [Hampshire] Dead server </div><div>
</div>On 20/02/15 22:12, Samuel Penn wrote:
> On Friday 20 Feb 2015 21:27:46 Leo wrote:
>> I noticed today that my headless server stopped responding so I had to
>> reboot it. However all I can tell is that the logs appear to have
>> stopped recording anything from about mid-afternoon yesterday. Is there
>> anywhere else I can look to see what might have happened?
>
> I was getting that recently. Turned out to be disc errors.
>

So, a very late follow up to this: I tested the disks using fsck and 
smart, and found no problems. So I ignored the problem. However, in the 
last week it has since frozen twice again. So I ram memtest. The first 
time it reported some errors. The second time, it didn't report any.

Has anyone had any experience of failed memory: is it possible to pass 
and fail, or once failing will it always fail?

Thanks,
Leo

PS I'm now running an mprime torture test to see what that throws up.

-- 
Please post to: [email protected]
Web Interface: https://mailman.lug.org.uk/mailman/listinfo/hampshire
LUG URL: http://www.hantslug.org.uk
--------------------------------------------------------------
-- 
Please post to: [email protected]
Web Interface: https://mailman.lug.org.uk/mailman/listinfo/hampshire
LUG URL: http://www.hantslug.org.uk
--------------------------------------------------------------

Reply via email to