From owner-freebsd-fs@FreeBSD.ORG Mon Dec 1 20:25:28 2014 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 69D16F4A for ; Mon, 1 Dec 2014 20:25:28 +0000 (UTC) Received: from mwork.nabble.com (mwork.nabble.com [162.253.133.43]) by mx1.freebsd.org (Postfix) with ESMTP id 44EA91F3 for ; Mon, 1 Dec 2014 20:25:27 +0000 (UTC) Received: from msam.nabble.com (unknown [162.253.133.85]) by mwork.nabble.com (Postfix) with ESMTP id 21C04BCD73F for ; Mon, 1 Dec 2014 07:36:21 -0800 (PST) Date: Mon, 1 Dec 2014 08:36:20 -0700 (MST) From: Dmitriy Makarov To: freebsd-fs@freebsd.org Message-ID: <1417448180496-5969924.post@n5.nabble.com> In-Reply-To: <547C6F59.9090106@internetx.com> References: <1417438604.143909513.k6b3b33f@frv41.fwdcdn.com> <547C6F59.9090106@internetx.com> Subject: Re: Irregular disk IO and poor performance (possibly after reading a lot of data from pool) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.18-1 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 01 Dec 2014 20:25:28 -0000 InterNetX - Juergen Gotteswinter-2 wrote > one/more disks broken or tend to die soon? Nope, at least does not seem like that. Smarts are OK for all drives: OK - HDD S.M.A.R.T health: src=0, rsc=0, rec=0, cps=0, ou=0, HEALTH_STATUS=PASSED Drives are not so old either ~1year. I doubt it has anything to do with the drives for two reasons: * paired devices in the mirror behave pretty much exactly * different pairs have 0 wtire I/O at different points of time and it's not like seconds, it is a matter of minutes. It seems as if ZFS decides to ignore some mirror devices for unknown reason. Randomly. Sometimes just 2-3 of 18 are idle. Sometimes it's 17 of 18! Nedless to say in that moment, that one poor mirror is doing all the hard work by itself. At which point system almost freezes, all I/O bound processes are blocked waiting for disk. -- View this message in context: http://freebsd.1045724.n5.nabble.com/Irregular-disk-IO-and-poor-performance-possibly-after-reading-a-lot-of-data-from-pool-tp5969896p5969924.html Sent from the freebsd-fs mailing list archive at Nabble.com.