From owner-freebsd-fs@FreeBSD.ORG Fri Mar 28 20:40:08 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 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 1EDA8ED4 for ; Fri, 28 Mar 2014 20:40:08 +0000 (UTC) Received: from woozle.rinet.ru (woozle.rinet.ru [195.54.192.68]) (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 AACD0F58 for ; Fri, 28 Mar 2014 20:40:07 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by woozle.rinet.ru (8.14.5/8.14.5) with ESMTP id s2SKe4nV047536; Sat, 29 Mar 2014 00:40:04 +0400 (MSK) (envelope-from marck@rinet.ru) Date: Sat, 29 Mar 2014 00:40:04 +0400 (MSK) From: Dmitry Morozovsky To: Joar Jegleim Subject: Re: zfs l2arc warmup In-Reply-To: Message-ID: References: <20140328005911.GA30665@neutralgood.org> User-Agent: Alpine 2.00 (BSF 1167 2008-08-23) X-NCC-RegID: ru.rinet X-OpenPGP-Key-ID: 6B691B03 MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.4.3 (woozle.rinet.ru [0.0.0.0]); Sat, 29 Mar 2014 00:40:05 +0400 (MSK) Cc: "freebsd-fs@freebsd.org" X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.17 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 28 Mar 2014 20:40:08 -0000 On Fri, 28 Mar 2014, Joar Jegleim wrote: [snip most of] > > Have you measured to see if, or do you otherwise know for sure, that you > > really do need a ZIL? I suggest not adding a ZIL unless you are certain > > you need it. > Yes, I only recently realized that too, and I'm really not sure if a > zil is required. > Some small portion of files (som hundre MB's) are served over nfs from > the same server, if I understand it right a zil will help for nfs > stuff (?) , but I'm not sure if it's any gain of having a zil today. > On the other hand, a zil doesn't have to be big, I can simply buy a > 128GB ssd which are cheap today . Please don't forget that, unlike L2ARC, if you lost ZIL during sync write, you're effectively lost the pool. Hence, you have two points: - have ZIL on a enterprise-grade SLC SSD (aircraft-grade prices ;P) - allocate mirrored ZIL from fraction (rule of thumb if I'm not mistaken was "get all of write performance of your low-level disks per 1 second, double it, and it will be size of your ZIL) of your existing otherwise used for L2ARC SSDs We (by all means not at your read pressure) used the second approach, like the following: pool: br state: ONLINE scan: resilvered 13.0G in 0h3m with 0 errors on Sun Aug 18 19:52:20 2013 config: NAME STATE READ WRITE CKSUM br ONLINE 0 0 0 mirror-0 ONLINE 0 0 0 gpt/br0 ONLINE 0 0 0 gpt/br4 ONLINE 0 0 0 mirror-1 ONLINE 0 0 0 gpt/br1 ONLINE 0 0 0 gpt/br5 ONLINE 0 0 0 mirror-2 ONLINE 0 0 0 gpt/br2 ONLINE 0 0 0 gpt/br6 ONLINE 0 0 0 mirror-3 ONLINE 0 0 0 gpt/br3 ONLINE 0 0 0 gpt/br7 ONLINE 0 0 0 logs mirror-4 ONLINE 0 0 0 gpt/br-zil0 ONLINE 0 0 0 gpt/br-zil1 ONLINE 0 0 0 cache gpt/br-l2arc0 ONLINE 0 0 0 gpt/br-l2arc1 ONLINE 0 0 0 where logs/cache are like root@briareus:~# gpart show -l da9 => 34 234441581 da9 GPT (111G) 34 2014 - free - (1M) 2048 16777216 1 br-zil0 (8.0G) 16779264 217661440 2 br-l2arc0 (103G) 234440704 911 - free - (455k) (this is our main PostgreSQL server, with 8 SASes and 2*Intel3500 SSDs) -- Sincerely, D.Marck [DM5020, MCK-RIPE, DM3-RIPN] [ FreeBSD committer: marck@FreeBSD.org ] ------------------------------------------------------------------------ *** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- marck@rinet.ru *** ------------------------------------------------------------------------