Wondering if anyone can shed some light here...

I have a RH8 box which has been running happily for about a week when I set
it up (Linux n00b here). Managed to get Samba and iptables and all that
working (great learning xp) so everything works great over the network..

BUT

Periodically, the server freezes completely. No error messages, nothing, and
all that's left to do is reboot. No explanation. It often happens when I'm
moving large files or doing several things on the server at once, but when I
leave it overnight, I'll come back down in the morning and it'll have
crashed.

It's a fairly dated box (300mhz, 96ram) but could run Win2k advanced server
fine, which leads me to think that it could be a hardware-dying issue...
most of the components were find-me-around-the-office parts that weren't
being used and I suspect have a few years and a lot of use behind them.

Could RAM be an issue (ie. dodgy sticks)??

Any ideas??

TIA,

--

George Hewitt <george@nospamgeorgehewitt.co.uk> wrote:
> I have a RH8 box which has been running happily for about a week when I set
[...]
> Periodically, the server freezes completely. No error messages, nothing, and

Same here, except RH 9, I do sometimes get error messages about
kswapd, but not always.

> all that's left to do is reboot. No explanation. It often happens when I'm
> moving large files or doing several things on the server at once, but when I
> leave it overnight, I'll come back down in the morning and it'll have
> crashed.

Mine *only* happens when I'm moving large directories over the network
from the older server to my new one, dirs over 5G.

I have found a solution. Sort of.

If I disable my second swap partition (the hint was the kswapd
message), all is well.

But when I run mkswap -c on it, all seems well. I wrote a perl script
to write patterns of 1's and 0's, then read them back, never any
errors.

Next I'm going to try the "data lifguard" disk that came with the
drive. I think it's destructive, but I have mirrored drives, so I
should be able to rebuild afterwards.

> Could RAM be an issue (ie. dodgy sticks)??

Just in case, I ran memtest86 for 25 hours on mine, no errors.

--
Jim Buchanan jbuchana@kokomo1.net jbuchana@buchanan1.net
=================== http://www.buchanan1.net/ ==========================
I'm gonna hit the highway like a battering ram, on a silver black
phantom bike, when the metal is hot and the engine is hungry and we're
all about to see the light... -Meatloaf
================= Visit: http://www.thehungersite.com ==================

Thanks for the comments.

I went drastic earlier this afternoon. The server was one of two (very)
cheap boxes I bought a while back. Both boxes were identical. The results
suggest it was a knackered mobo or dodgy/not enough RAM.

What I did was to move the server's HDs and processor slotcard to the new
machine (it wouldn't POST with the other machine's proc for some reason).
Booted without any trouble and now been running it all afternoon no
problems. I moved two very large DVD isos to the server and back again just
to check and it's all ok. Runs faster now as well since this box has 192mb
ram rather than the 96 in the other.

Hope you can find a solution to yours.

--

<jbuchana@buchanan1.net> wrote in message
news:TugKa.14310$Bg.9513@rwcrnsc54...
> George Hewitt <george@nospamgeorgehewitt.co.uk> wrote:
> > I have a RH8 box which has been running happily for about a week when I
set
> [...]
> > Periodically, the server freezes completely. No error messages, nothing,
and
>
> Same here, except RH 9, I do sometimes get error messages about
> kswapd, but not always.
>
> > all that's left to do is reboot. No explanation. It often happens when
I'm
> > moving large files or doing several things on the server at once, but
when I
> > leave it overnight, I'll come back down in the morning and it'll have
> > crashed.
>
> Mine *only* happens when I'm moving large directories over the network
> from the older server to my new one, dirs over 5G.
>
> I have found a solution. Sort of.
>
> If I disable my second swap partition (the hint was the kswapd
> message), all is well.
>
> But when I run mkswap -c on it, all seems well. I wrote a perl script
> to write patterns of 1's and 0's, then read them back, never any
> errors.
>
> Next I'm going to try the "data lifguard" disk that came with the
> drive. I think it's destructive, but I have mirrored drives, so I
> should be able to rebuild afterwards.
>
> > Could RAM be an issue (ie. dodgy sticks)??
>
> Just in case, I ran memtest86 for 25 hours on mine, no errors.
>
> --
> Jim Buchanan jbuchana@kokomo1.net jbuchana@buchanan1.net
> =================== http://www.buchanan1.net/ ==========================
> I'm gonna hit the highway like a battering ram, on a silver black
> phantom bike, when the metal is hot and the engine is hungry and we're
> all about to see the light... -Meatloaf
> ================= Visit: http://www.thehungersite.com ==================