On Thu, Jul 29, 2004 at 12:51:48AM +0200, Peter Bex wrote:
> Hi there,
> 
> Somehow I ran into the following problem today:
> Whenever I start a `big' program like gcc, it dies hard and I see messages like:
> 
> UVM: pid 10134 (cc1), uid 0 killed: out of swap
> 
> in my dmesg.  I have more than enough swap:
> 
> $ swapctl -l
> Device      512-blocks     Used    Avail Capacity  Priority
> /dev/wd1b      1050273    26240  1024033     2%    0
> 
> And the swap partition lives on a harddisk that is something like 4 weeks old.
> 
> What should I do to investigate the cause of this problem?  I won't have
> physical access to the machine, so I can't do anything to the hardware.
> 
> I can't use my machine properly anymore, so I'd *really* like to fix this
> asap.
> 
> Some info:
> 
> NetBSD 2.0_BETA (FROHIKE) #1: Tue Jul  6 14:42:10 CEST 2004
> 
> total memory = 65084 KB
> avail memory = 60912 KB
> cpu0: Intel Pentium II (686-class), 350.81 MHz, id 0x652
> 
> The first harddisk is old and shows some errors in dmesg:
> 
> wd0: transfer error, downgrading to Ultra-DMA mode 1
> wd0(piixide0:0:0): using PIO mode 4, Ultra-DMA mode 1 (using DMA data transfers)
> wd0a: error writing fsbn 1960416 of 1960416-1960431 (wd0 bn 1960479; cn 2074
> tn 8 sn 45), retrying
> wd0: (aborted command, interface CRC error)
> wd0: soft error (corrected)
> 
> Could this somehow have an effect with regards to the swap problem?

Probably not. I suspect this is more a problem with the buffer cache.
Try reducing vm.filemax and vm.bufcache, and maybe bumping vm.anonmin
and vm.execmin

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--