From owner-freebsd-current@FreeBSD.ORG Fri Dec 18 10:15:26 2009 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 1639C106568B for ; Fri, 18 Dec 2009 10:15:26 +0000 (UTC) (envelope-from emikulic@gmail.com) Received: from ipmail03.adl6.internode.on.net (ipmail03.adl6.internode.on.net [203.16.214.141]) by mx1.freebsd.org (Postfix) with ESMTP id 866348FC0A for ; Fri, 18 Dec 2009 10:15:25 +0000 (UTC) X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: ApsEAD/kKkuWZZrw/2dsb2JhbACSFrJ2hyKIaYQtBA Received: from ppp154-240.static.internode.on.net ([150.101.154.240]) by ipmail03.adl6.internode.on.net with ESMTP; 18 Dec 2009 20:45:23 +1030 Received: by ppp154-240.static.internode.on.net (Poo-fix, from userid 1001) id 6EAB65C46; Fri, 18 Dec 2009 21:15:22 +1100 (EST) Date: Fri, 18 Dec 2009 21:15:22 +1100 From: Emil Mikulic To: Alexander Zagrebin Message-ID: <20091218101522.GD71197@dmr.ath.cx> References: <398F182D4FC14D43A1CFD89C4D69817C@vosz.local> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <398F182D4FC14D43A1CFD89C4D69817C@vosz.local> User-Agent: Mutt/1.5.20 (2009-06-14) Cc: freebsd-current@freebsd.org Subject: Re: ZFS and UFS on the same box. Cache memory allocation issue. X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 18 Dec 2009 10:15:26 -0000 On Fri, Dec 18, 2009 at 11:49:47AM +0300, Alexander Zagrebin wrote: > As result UFS returns memory to "free" pool and ZFS can allocate it for ARC. > It significantly increases ZFS performance. > > Is there an another workaround? I use: (cd /somefs && sudo umount /somefs) The umount fails, but it does clear the buffer cache! Memory reported as "inactive" by top decreases. I find this to be a really handy trick when benchmarking I/O. ;) Unfortunately, it doesn't work on the root filesystem. --Emil