From owner-freebsd-questions@FreeBSD.ORG Fri Sep 29 16:55:40 2006 Return-Path: X-Original-To: freebsd-questions@freebsd.org Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id DB0DA16A4C9 for ; Fri, 29 Sep 2006 16:55:40 +0000 (UTC) (envelope-from snagit@cbpratt.prohosting.com) Received: from n054.sc0.cp.net (smtpout1107.sc0.he.tucows.com [64.97.144.107]) by mx1.FreeBSD.org (Postfix) with ESMTP id A655943D53 for ; Fri, 29 Sep 2006 16:48:14 +0000 (GMT) (envelope-from snagit@cbpratt.prohosting.com) Received: from [192.168.1.100] (67.47.213.85) by n054.sc0.cp.net (7.2.069.1) (authenticated as eagletree@hughes.net) id 451CC919000175C8 for freebsd-questions@freebsd.org; Fri, 29 Sep 2006 16:47:43 +0000 Mime-Version: 1.0 (Apple Message framework v752.2) In-Reply-To: <20060929160631.GE657@gizmo.acns.msu.edu> References: <20060929160631.GE657@gizmo.acns.msu.edu> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <569ECAEE-4AEF-431B-B724-E734F8EE44E2@cbpratt.prohosting.com> Content-Transfer-Encoding: 7bit From: Chris Date: Fri, 29 Sep 2006 09:47:31 -0700 To: FreeBSD Questions X-Mailer: Apple Mail (2.752.2) Subject: Re: Swap Size Importance? X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 29 Sep 2006 16:55:41 -0000 On Sep 29, 2006, at 9:06 AM, Jerry McAllister wrote: > On Fri, Sep 29, 2006 at 08:52:58AM -0700, Chris wrote: > >> >> Is there any shortfall in performance or reliability to running >> production with swap equal in size to the 8gb of system memory? I > > doesn't matter much. But, if you run enough to actually cause > paging - which goes to swap space - then it becomes an issue. Also, > I am assuming that real paging of active processes is death to that server anyway and means something else has to be throttled back with tuning of network bufs, apache or mysql. Same for crash dumps, can't run a server that is taking dumps or you lose your traffic. > I think some things that get pulled to execute often can get left > in swap space and accessed more quickly that all the way from main > disk each time. eg the system keeps track of what it has in swap and > it is more efficient to read from swap - less overhead. But someone This is the part that concerned me. If one views a top on well running system and sees no swapping, I wanted to make certain there is no magic going on behind the scenes where processes have been mapped to swap in such a way that I could be currently benefitting from swap being higher than actual and not know it. If top is an accurate read on whether the system has placed high use processes in swap then it would suggest the first post is correct, and a memory rich system, where you configure to never exceed real memory, wastes that storage taken in swap. For expensive drives, given the sizes we use in RAM now, it's hard to justify. In the case of attempting this raid-5 configuration, it equates to the loss of 24G in scsi storage. I will run with 8g on the system drive. Thank you very much for the responses.