From dillon@flea.best.net  Tue Dec  2 17:14:16 1997
Received: from flea.best.net (root@flea.best.net [206.184.139.131])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id RAA22342
          for <FreeBSD-gnats-submit@freebsd.org>; Tue, 2 Dec 1997 17:14:16 -0800 (PST)
          (envelope-from dillon@flea.best.net)
Received: (from dillon@localhost) by flea.best.net (8.8.7/8.7.3) id RAA17646; Tue, 2 Dec 1997 17:14:06 -0800 (PST)
Message-Id: <199712030114.RAA17646@flea.best.net>
Date: Tue, 2 Dec 1997 17:14:06 -0800 (PST)
From: Matt Dillon <dillon@best.net>
Reply-To: dillon@best.net
To: FreeBSD-gnats-submit@freebsd.org
Subject: Idle-swapout / swapper ineffective
X-Send-Pr-Version: 3.2

>Number:         5197
>Category:       kern
>Synopsis:       The swapper is ineffective
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-bugs
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Dec  2 17:20:00 PST 1997
>Closed-Date:    Fri May 8 13:45:05 PDT 1998
>Last-Modified:  Fri May  8 13:49:05 PDT 1998
>Originator:     Matt Dillon
>Release:        FreeBSD 2.2.5-STABLE i386
>Organization:
Best Internet Communications
>Environment:

	FreeBSD 2.2.5

>Description:

	The swapper (verses the pager) is somewhat ineffective, activating
	only during a severe memory shortfall.    The only current way to 
	tune swap operation is via vm.v_free_target, but setting it high 
	enough to cause swapping tends to result in too much swapping.  i.e.
	the parameter is too sensitive when it comes to the swapping
	code (verses the paging code).

>How-To-Repeat:

	n/a

>Fix:
	
	It would be useful to have the vm_pageout code swap out processes
	idle over N seconds, where N is programmable via a sysctl.  In fact,
	if this were in place I would almost consider removing the 
	'severe memory shortfall activates swapper' code.

>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->closed 
State-Changed-By: phk 
State-Changed-When: Fri May 8 13:45:05 PDT 1998 
State-Changed-Why:  
Doing what you suggest would prevent the swapper from acting as 
last defense against low-mem conditions.  I belive the future of 
the swapper is generally uncertain with a tendency to "kill it!" 
being mumbled.  (Look for NO_SWAPPING in the -current vmcode). 
>Unformatted:
