Re: [Hampshire] Dead server

>     On 14/05/15 09:40, Gordon Scott wrote:

>         I'd go along with that.

>         The ones that normally go are the electrolytic types ..
>         aluminium cans
>         with black(usually) printing. The electrolyte is a liquid and
>         tends to
>         dry out over a number of years use in a warm environment. Swelling,
>         (usually of the flat top), discolouration, oozing electrolyte.

>         The next most likely candidates are tantalum capacitors, which
>         tend to
>         be little black rectangular block. When they fail, they tend to
>         blow a
>         corner off of the moulding, or sometimes just a small hole/crater.

>         Most of the rest will be ceramics, which are usually trouble-free.

>         Gordon.

>     So I've had a look at the capacitors, and I can't see any that look
>     broken. I've also done some more investigation and found the
>     following: if the computer locks up and I then run memtest on reboot
>     it finds errors in the same memory locations each time. However if I
>     reboot cleanly it doesn't find errors. The fact it finds them in the
>     same locations would indicate to me that it's a memory problem.
>     However, I also ran the mprime torture test, and that failed on both
>     the memory intensive test, and the test that doesn't use much
>     memory. Which would tend to indicate that it's not a memory problem.

>     I'm now trying a kernel parameter that should stop it using the
>     "bad" memory to see if that fixes it...

>     Leo

>     --
>     Please post to: Hampshire@???
>     <mailto:Hampshire@mailman.lug.org.uk>
>     Web Interface: https://mailman.lug.org.uk/mailman/listinfo/hampshire
>     LUG URL: http://www.hantslug.org.uk
>     --------------------------------------------------------------