From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: with ECARTIS (v1.0.0; list xfs); Thu, 29 Nov 2007 23:33:10 -0800 (PST) Received: from mars.puhasoft.hu (mars.puhasoft.hu [212.108.197.33]) by oss.sgi.com (8.12.11.20060308/8.12.10/SuSE Linux 0.7) with ESMTP id lAU7X3Gj029677 for ; Thu, 29 Nov 2007 23:33:06 -0800 Message-ID: <474FB90C.4060905@tsabi.hu> Date: Fri, 30 Nov 2007 08:17:32 +0100 From: =?UTF-8?B?VMOzdGggQ3NhYmE=?= MIME-Version: 1.0 Subject: Re: XFS performance problems on Linux x86_64 References: <474C8A05.3020604@e-626.net> <20071127220536.GL119954183@sgi.com> <20071130045808.GK119954183@sgi.com> In-Reply-To: <20071130045808.GK119954183@sgi.com> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sender: xfs-bounce@oss.sgi.com Errors-to: xfs-bounce@oss.sgi.com List-Id: xfs To: David Chinner Cc: Bernd Schubert , linux-xfs@oss.sgi.com Hello list, I tried this parameters, and got this results with bonnie++. As i think there isnt any speedup with this parameteres, or i am doing something wrong? tsabi oldbck ~ # uname -a Linux oldbck 2.6.23-gentoo-r2-uk-01 #1 SMP Tue Nov 20 03:43:04 CET 2007 x86_64 Intel(R) Xeon(TM) CPU 2.80GHz GenuineIntel GNU/Linux test 1: oldbck mnt # mkfs.xfs -i size=512 -f /dev/md5 meta-data=/dev/md5 isize=512 agcount=32, agsize=4513008 blks = sectsz=512 attr=0 data = bsize=4096 blocks=144416192, imaxpct=25 = sunit=16 swidth=64 blks, unwritten=1 naming =version 2 bsize=4096 log =internal log bsize=4096 blocks=32768, version=1 = sectsz=512 sunit=0 blks, lazy-count=0 realtime =none extsz=262144 blocks=0, rtextents=0 oldbck mnt # mount /dev/md5 /mnt/data oldbck mnt # bonnie++ -u root -d /mnt/data Version 1.93c ------Sequential Output------ --Sequential Input- --Random- Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP oldbck 4G 522 99 235160 50 86692 17 1040 96 241681 22 457.9 9 Latency 15620us 205ms 119ms 100ms 50727us 79657us Version 1.93c ------Sequential Create------ --------Random Create-------- oldbck -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete-- files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP 16 11000 77 +++++ +++ 13765 82 11977 83 +++++ +++ 11049 77 Latency 66657us 49us 76936us 68300us 18us 72243us 1.93c,1.93c,oldbck,1,1196392150,4G,,522,99,235160,50,86692,17,1040,96,241681,22,457.9,9,16,,,,,11000,77,+++++,+++,13765,82,11977,83,+++++,+++,11049,77,15620us,205ms,119ms,100ms,50727us,79657us,66657us,49us,76936us,68300us,18us,72243us test 2: oldbck mnt # mkfs.xfs -f -l lazy-count=1,version=2,size=128m -i attr=2,size=512 -d agcount=4 /dev/md5 meta-data=/dev/md5 isize=512 agcount=4, agsize=36104048 blks = sectsz=512 attr=2 data = bsize=4096 blocks=144416192, imaxpct=25 = sunit=16 swidth=64 blks, unwritten=1 naming =version 2 bsize=4096 log =internal log bsize=4096 blocks=32768, version=2 = sectsz=512 sunit=16 blks, lazy-count=1 realtime =none extsz=262144 blocks=0, rtextents=0 oldbck mnt # mount -o logbsize=256k,nobarrier /dev/md5 /mnt/data oldbck mnt # bonnie++ -u root -d /mnt/data Version 1.93c ------Sequential Output------ --Sequential Input- --Random- Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP oldbck 4G 523 99 237016 44 87202 17 1040 96 245389 22 446.5 7 Latency 15531us 184ms 133ms 105ms 11835us 85541us Version 1.93c ------Sequential Create------ --------Random Create-------- oldbck -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete-- files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP 16 12772 77 +++++ +++ 14759 74 13499 81 +++++ +++ 11617 70 Latency 86835us 56us 79869us 79967us 24us 95818us 1.93c,1.93c,oldbck,1,1196391889,4G,,523,99,237016,44,87202,17,1040,96,245389,22,446.5,7,16,,,,,12772,77,+++++,+++,14759,74,13499,81,+++++,+++,11617,70,15531us,184ms,133ms,105ms,11835us,85541us,86835us,56us,79869us,79967us,24us,95818us test 3: oldbck mnt # mkfs.ext3 /dev/md5 mke2fs 1.40.2 (12-Jul-2007) Filesystem label= OS type: Linux Block size=4096 (log=2) Fragment size=4096 (log=2) 72220672 inodes, 144416192 blocks 7220809 blocks (5.00%) reserved for the super user First data block=0 Maximum filesystem blocks=4294967296 4408 block groups 32768 blocks per group, 32768 fragments per group 16384 inodes per group Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208, 4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968, 102400000 Writing inode tables: done Creating journal (32768 blocks): done Writing superblocks and filesystem accounting information: done This filesystem will be automatically checked every 32 mounts or 180 days, whichever comes first. Use tune2fs -c or -i to override. oldbck mnt # mount /dev/md5 /mnt/data oldbck mnt # oldbck mnt # bonnie++ -u root -d /mnt/data Version 1.93c ------Sequential Output------ --Sequential Input- --Random- Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP oldbck 4G 326 99 211619 80 91680 23 1341 96 237493 21 495.5 9 Latency 32953us 220ms 1599ms 62374us 59342us 472ms Version 1.93c ------Sequential Create------ --------Random Create-------- oldbck -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete-- files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP 16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ Latency 7677us 165us 252us 17010us 13us 243us 1.93c,1.93c,oldbck,1,1196389368,4G,,326,99,211619,80,91680,23,1341,96,237493,21,495.5,9,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,32953us,220ms,1599ms,62374us,59342us,472ms,7677us,165us,252us,17010us,13us,243us David Chinner írta: > On Wed, Nov 28, 2007 at 12:13:57AM +0100, Bernd Schubert wrote: >> Hello David, >> >> David Chinner wrote: >>> # mkfs.xfs -f -l lazy-count=1,version=2,size=128m -i attr=2 -d agcount=4 >>> # mount -o logbsize=256k >> thanks, I was also going to ask which are optimal parameters. Just didn't >> have the time yet :) >> Any idea when these options will be default? > > They should already be the defaults in the current CVS tree. ;) > > Cheers, > > Dave.