lkml.org 
[lkml]   [2003]   [Aug]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: VM: killing process amavis
Date
On Wednesday 13 August 2003 15:40, Coen Rosdorff wrote:
> On Wed, 13 Aug 2003, Hugh Dickins wrote:
> > It really would be worth giving memtest86 a good long run.
> >
> > 02000000 looks very much like a single-bit memory error,
> > and swap_free is exactly where such errors often show up.
>
> I had the same problem before on the previous server. Running memtest for
> 19 days didn't showed any memory problems.
>
> After replacing the motherboard cpu and ram, now I have the same problem.

I had a system once that looked very much like it had bad ram, but it turned
out to have a bad hard drive controller, which showed up paging stuff into
memory from disk (ala exec, sometimes), and in bringing stuff back in from
swap. (The kernel almost never went bye-bye, because it never swapped out,
you see...)

Caused the weirdest problems in Myth II, among other things...

> So the problem moved from 00000100 to 02000000
>
> The networkcards and the 3ware raid controler moved form the old to the
> new box. Could one of them be the problem?
>
> I am running out of options.

Check the raid controller. Especially if you're swapping through the raid
controller. I found out what was wrong with the other system by copying big
tarballs through the network and verifying them.

Try this:

1) Copy a tarball to the remote system and confirm that it came out OK just
coming across the network.

cat enormous.tgz | ssh othersystem "tar tvz"

2) Now copy the tarball to the remote machine's disk, and test that the copy
on disk is good.

cat enormous.tgz | ssh othersystem "cat > temp.tgz; tar tvzf temp.tgz"

Of course using a tarball that's bigger than your ram, so it actually does
have to write it out to disk and read it back in again. Using ssh provides a
little bit of a CPU load, and of course the network is providing a competing
source of interrupts. (You could also run contest in the background or some
such to really beat the system to death...)

Rob


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:47    [W:0.062 / U:0.552 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site