From owner-freebsd-fs@FreeBSD.ORG Sat Apr 27 08:06:31 2013 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by hub.freebsd.org (Postfix) with ESMTP id AD3F3BFF; Sat, 27 Apr 2013 08:06:31 +0000 (UTC) (envelope-from daniel@digsys.bg) Received: from smtp-sofia.digsys.bg (smtp-sofia.digsys.bg [193.68.21.123]) by mx1.freebsd.org (Postfix) with ESMTP id 241B11227; Sat, 27 Apr 2013 08:06:30 +0000 (UTC) Received: from digsys200-136.pip.digsys.bg (digsys200-136.pip.digsys.bg [193.68.136.200]) (authenticated bits=0) by smtp-sofia.digsys.bg (8.14.6/8.14.6) with ESMTP id r3R7n99f053438 (version=TLSv1/SSLv3 cipher=AES128-SHA bits=128 verify=NO); Sat, 27 Apr 2013 10:49:10 +0300 (EEST) (envelope-from daniel@digsys.bg) Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 (Mac OS X Mail 6.3 \(1503\)) Subject: Re: Strange slowdown when cache devices enabled in ZFS From: Daniel Kalchev In-Reply-To: Date: Sat, 27 Apr 2013 10:49:09 +0300 Content-Transfer-Encoding: quoted-printable Message-Id: <6F1DCA37-9159-47F2-9ABB-FDEC41665611@digsys.bg> References: <51430744.6020004@FreeBSD.org> <51487CE1.5090703@FreeBSD.org> To: Freddie Cash X-Mailer: Apple Mail (2.1503) Cc: FreeBSD Filesystems , Andriy Gapon X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 27 Apr 2013 08:06:31 -0000 On Apr 25, 2013, at 8:44 PM, Freddie Cash wrote: > Since these pools all have dedupe enabled, is it possible that the = L2ARC > feed thread searching the ARC for stuff to write to L2 is = starving/blocking > reads of the DDT from the L2ARC thus preventing the pool from writing = any > new data? This seems very logical and would probably explain it. If your SSD is = fast enough, there should be no reason to keep the default. I would = guess that the "typical" (Sun's?) L2ARC SSD was very slow at writing and = letting it do reads at the same time was awful. But -- this might be a wrong track anyway -- do your dtraces :-) Daniel=