[comp.sys.sgi] Swap space disk striping under 3.3 wanted

doelz@urz.unibas.ch (06/29/90)

Release 3.3. supposedly supports 'striping' of disks on efs systems. 
I'd like to stripe my swap space (/debug) which is of type 'rawdata'
in the fx. Any pointers welcome. 

- Reinhard 

************************************************************************
    Dr. Reinhard Doelz            *     EAN     doelz@urz.unibas.ch
      Biocomputing                *     DECNET  48130::doelz
Biozentrum der Universitaet       *     X25     psi%46211142::embnet
   Klingelbergstrasse 70          *     FAX     x41 61 256760
     CH 4056 Basel                *     TEL     x41 61 253880 ext 888
************************************************************************

archer@elysium.esd.sgi.com (Archer Sully) (06/29/90)

In article <1990Jun29.083658.807@urz.unibas.ch> doelz@urz.unibas.ch writes:

>Release 3.3. supposedly supports 'striping' of disks on efs systems. 
>I'd like to stripe my swap space (/debug) which is of type 'rawdata'
>in the fx. Any pointers welcome. 

striping the swap space is unnecessary, as the system already handles
multiple swap devices that can even live on different controllers.
See swap(1m).

--
Archer Sully 	      |  Ask not what you can do for you country,
(archer@esd.sgi.com)  |  But what your country's been doing to you.
                      |                    -- The Avengers

jwag@moose.asd.sgi.com (Chris Wagner) (06/29/90)

Swap space, while still on raw partitions in 3.3 have always been
interleaved across as
many swap devices as a re configured. So if you have 4 swap areas, swap
will be interleaved
on a per request  basis across all 4 devices.

CHris Wagner

ktureski@alias.uucp (Kevin Tureski) (07/01/90)

In article <9737@odin.corp.sgi.com> archer@elysium.esd.sgi.com (Archer Sully) writes:
>
>striping the swap space is unnecessary, as the system already handles
>multiple swap devices that can even live on different controllers.
>See swap(1m).

A request from ages ago is whether you can (or will be able to in some future
release) swap on devices on *another machine's* controller ...

I personally don't care if it's as slow as molasses, since it may be the
ONLY way I can run a mega-process. Consider the following situation:

I've got a machine with a 2-drive controller with, say, a pair of 780's.
One has root, just enough for /usr to be comfortable, and the rest (500Mb)
dedicated to swap.

The other disk is configured as one large filesystem full of user files.
It's chock full of everything that's required to run this large process.

The process requires 600Mb swap. I've only got 500. And can't get any more,
no way, no how on that machine without getting a 4-drive controller and
stealing a disk off of another machine.

But if I could just umount that other "extra" disk for a while and dedicate
it's swap space to that other machine, I'd get the job done without the pain
and anguish that I've personally gone through in the past.


Kevin Tureski
Director, Systems Integration
Alias Research Inc.

ktureski%alias@csri.utoronto.ca		ktureski@alias.uucp