From owner-freebsd-stable@FreeBSD.ORG Thu Sep 11 10:46:54 2014 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 42B7EF52; Thu, 11 Sep 2014 10:46:54 +0000 (UTC) Received: from constantine.ingresso.co.uk (constantine.ingresso.co.uk [IPv6:2a02:b90:3002:e550::3]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 0A019E0; Thu, 11 Sep 2014 10:46:54 +0000 (UTC) Received: from dilbert.london-internal.ingresso.co.uk ([10.64.50.6] helo=dilbert.ingresso.co.uk) by constantine.ingresso.co.uk with esmtps (TLSv1:DHE-RSA-AES256-SHA:256) (Exim 4.82 (FreeBSD)) (envelope-from ) id 1XS1tq-0005Nc-Fg; Thu, 11 Sep 2014 10:46:50 +0000 Received: from petefrench by dilbert.ingresso.co.uk with local (Exim 4.84 (FreeBSD)) (envelope-from ) id 1XS1tq-000IW6-DB; Thu, 11 Sep 2014 11:46:50 +0100 To: ari@ish.com.au, freebsd-stable@freebsd.org, petefrench@ingresso.co.uk, se@freebsd.org Subject: Re: getting to 4K disk blocks in ZFS In-Reply-To: <54117624.7020907@freebsd.org> Message-Id: From: Pete French Date: Thu, 11 Sep 2014 11:46:50 +0100 X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.18-1 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 11 Sep 2014 10:46:54 -0000 > There is no problem, if ashift covers more than 1 sector, except > that you waste some space. If ashift=12 is used with 512 byte > sectors, then all writes will be to 8 consecutive sectors. There > is no read-modify-write as in the opposite case (ashift=9 with 4K > sectors). But what if one of those 8 writes fails ? ZFS understands that part of a write can fail then ? or does it think of it as 8 separate writes ? > But the amount of wasted space can be quite substantial. I have > read reports of some 8% less usable space with ashift=12 compared > to ashift=9, for an empty ZFS file system. And with lots of small > files, this will become worse, once the file system is filled. Yes, I have migrated a lot of filesystems over to 4k, and there is quite a lot of extra space taken up by the same amount of data - even wth lz4 compression enabled, which I do on most systems. However the extra performance is well worth it. Only reason I am asking is that I am thinking of migrating a zpool which only has one 4k drive in it, plsy 3 512byte ones. Just want to make sure that it is not a bad idea to do this. -pete.