Mailinglist Archive: opensuse (3666 mails)

< Previous Next >
Re: [SLE] Maximum swap size
  • From: Paul Hewlett <hewlett@xxxxxxxxxxxxxxx>
  • Date: Mon, 7 Mar 2005 09:53:48 +0200
  • Message-id: <200503070953.48759.hewlett@xxxxxxxxxxxxxxx>
On Monday 07 March 2005 01:10, Hans Witvliet wrote:
> On Sun, 2005-03-06 at 19:44, James Knott wrote:
> > This was more of a curiousity type question. Also, IIRC, you can have
> > multiple swap partitions, so terabyte swap space is now a possibility,
> > though I'd imagine exabyte swap is still a while away.
> >
> > Incidentally, it wasn't all that long ago, that a fried was bragging
> > that he had a total (4 or 5 drives) of 1 GB disk space on his OS/2
> > system!
>
> I allready though so,
> For most "normal" situations, sysadmins consider swap nowadays more as a
> protection meganism against kernel panics.
>
> Used to be different. I recall the days (on a dec vax/vms 780) and
> several hundred people working on it, and on piek moments during the
> week, most of us were watching which users were swapped out ;-))
> At that time "swapped/paged-out" was a fact of life. But since then,
> CPU, cache and mem-speed have rissen 1000-fold; while "background"
> -memory, or disks in general perhaps 10 or 20-fold.
> So the effect of a process being swapped out of main-memory is felt much
> more.

I had a customer once running a dual P3 with 256MB RAM on which our software
was running 24/7 monitoring a mine in Canada. Customer complained that he
could not load one of our interactive analysis programs - I dialled in only
to discover that the swap partition had been corrupted and that the system
had been running for 3 weeks without any swap at all (RH7.3). Rerunning
mkswap on the partition and reissuing the swapon command fixed the problem.
Note that this machine ran a well-defined set of processes which rarely
changed and thus swap was not needed.

Paul
--
Paul Hewlett (Linux #359543) Email:`echo az.oc.evitcaten@ttelweh | rev`
Tel: +27 21 852 8812 Cel: +27 72 719 2725 Fax: +27 86 672 0563
--

< Previous Next >
Follow Ups