Re: hardware errors

[Richard Hector <richard@walnut.gen.nz>]

On 10/06/14 23:24, Ralf Mardorf wrote:
> On Tue, 2014-06-10 at 23:07 +1200, Richard Hector wrote:
> > On 10/06/14 23:04, Ralf Mardorf wrote:
> > > On Tue, 2014-06-10 at 21:24 +1200, Richard Hector wrote:
> > > > On 09/06/14 11:35, Bzzzz wrote:
> > > > > On Mon, 09 Jun 2014 11:22:25 +1200 Richard Hector
> > > > > <richard@walnut.gen.nz> wrote:
> > > > >
> > > > > > I assume the RAM needs replacing - is it possible to figure
> > > > > > out which DIMM(s)?
> > > > >
> > > > > Install memtest86+ and boot on it, then leave at least 3
> > > > > complete cycles to run.
> > > >
> > > > Thanks.
> > > >
> > > > Have created a memtest86+ CD and will try it tomorrow evening
> > > > (need a scheduled time to take it down).
> > > >
> > > > Interestingly, there are no more errors logged for the last day
> > > > and a half ...
> > > >
> > > > Any guesses as to how long these 3 complete cycles will take?
> > > > It's a Sun Fire X2100 M2 (dual core opteron 1218, 2600MHz) with
> > > > 4G of RAM. I haven't run memtest for ages ...
> > >
> > > IIRC one complete standard test with my dual-core Athlon 2.1 GHz 4
> > > GiB RAM takes more than 1 hour. I guess in 1 day it does around 8
> > > complete tests, perhaps I run it just during the night in half of a
> > > day. I might be mistaken, but you should expect that you need to
> > > run it for several hours.
> >
> > Thanks. I'm not sure how long we can afford to leave the machine down;
> > hopefully the error will show up promptly. BTW - it will show an error
> > even if ECC corrects it, right?
>
> No ECC here. I don't know.
>
> I used StartPage and searched for "memtest ECC". It seems to be, that
> memetst isn't good to test ECC. The current version seems to provide
> very limited hardware, seemingly Intel only.

Yep. Halfway through the third pass; no errors yet. I'm not holding my
breath.
Any ideas on where to read up on those error messages, to figure out
what they actually mean?

Richard