Fixing the vm or merging rmap into the official tree?

Fixing the vm or merging rmap into the official tree?

[Roy Sigurd Karlsbakk]

Hi all

After weeks of testing, knocking my head against all sorts of objects,
trying out other potential OSes etc. etc. ad. infinitum, I got the hint of
using the rmap patch to fix my problems with reading multiple large files
at once (see prevois thread with subject "[BUG] Error reading multiple
large files").

Will this problem be addressed in 2.4 or perhaps 2.[56] ?

My testing shows that the current vm can't handle high/non-standards load
efficiently. Isn't this something that clearly should be addressed?

roy

--
Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA

Computers are like air conditioners.
They stop working when you open Windows.

Re: Fixing the vm or merging rmap into the official tree?

[Martin Knoblauch]

> Fixing the vm or merging rmap into the official tree?
> 
> From: Roy Sigurd Karlsbakk (roy@karlsbakk.net)
> Date: Thu Jan 10 2002 - 04:55:39 EST
> 
> 
> Hi all
> 
> After weeks of testing, knocking my head against all sorts of objects,
> trying out other potential OSes etc. etc. ad. infinitum, I got the hint of
> using the rmap patch to fix my problems with reading multiple large files
> at once (see prevois thread with subject "[BUG] Error reading multiple
> large files").
> 
> Will this problem be addressed in 2.4 or perhaps 2.[56] ?
> 
> My testing shows that the current vm can't handle high/non-standards load
> efficiently. Isn't this something that clearly should be addressed?
>

 Just my 2 EURO-ct on this. The VM in 2.4.x mainline is definitely
broken for a lot of loads, especially under high memory stress. This
needs to be adressed and fixed ASAP. An I think most people here agree.

 Now, the question is what is the correct fix. There is Andreas´s stuff,
which fixes some scenarios. There is Rick's stuff which fixes others.
There is even the small few liner patch to vmscan.c by M.v.Leuwen, which
fixes/reduces the swapout problem for *my* situation.

 But frankly speaking, I personally don't care which solution goes in,
but it should be done quick. Patching up mainline every time just to get
decent VM behaviour is unacceptable for serious use.

> roy
> 
> --
> Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA
------------------------^^^^^^^^^^

 poor guy :-)

Martin
-- 
------------------------------------------------------------------
Martin Knoblauch         |    email:  Martin.Knoblauch@TeraPort.de
TeraPort GmbH            |    Phone:  +49-89-510857-309
C+ITS                    |    Fax:    +49-89-510857-111
http://www.teraport.de   |    Mobile: +49-170-4904759