Date: Tue, 30 Jan 2007 15:00:47 -0500 From: Robert Huff <roberthuff@rcn.com> To: freebsd-current@freebsd.org Subject: Re: Interesting speed benchmarks Message-ID: <17855.41967.928998.142397@jerusalem.litteratus.org> In-Reply-To: <20070130075708.GB892@turion.vk2pj.dyndns.org> References: <20070126224352.GD927@turion.vk2pj.dyndns.org> <17854.27369.656331.328734@jerusalem.litteratus.org> <20070130075708.GB892@turion.vk2pj.dyndns.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Peter Jeremy writes: > >huff@>>dump 0 -D /tmp/DF -Lau -f /dev/null /usr >& /tmp/null_dump > > I'd drop the '-u' and '-D' and add a '-C'. When sizing the cache, > take into account that each dump subprocess (typically around 5) > will allocate that much RAM. huff@>>dump 0 -La -C 32 -f /dev/null /usr >& /tmp/null_dump huff@>>cat /tmp/null_dump DUMP: Date of this level 0 dump: Tue Jan 30 08:54:55 2007 DUMP: Date of last level 0 dump: the epoch DUMP: Dumping snapshot of /dev/da1s1d (/usr) to /dev/null DUMP: mapping (Pass I) [regular files] DUMP: Cache 32 MB, blocksize = 65536 DUMP: mapping (Pass II) [directories] DUMP: estimated 26869650 tape blocks. DUMP: dumping (Pass III) [directories] DUMP: 0.55% done, finished in 15:06 at Wed Jan 31 00:06:45 2007 DUMP: dumping (Pass IV) [regular files] DUMP: 3.12% done, finished in 5:10 at Tue Jan 30 14:15:50 2007 DUMP: 6.47% done, finished in 3:36 at Tue Jan 30 12:47:08 2007 DUMP: 96.97% done, finished in 0:06 at Tue Jan 30 12:26:50 2007 DUMP: 98.77% done, finished in 0:02 at Tue Jan 30 12:28:03 2007 DUMP: 99.99% done, finished soon DUMP: DUMP: 27162681 tape blocks on 1 volume DUMP: finished in 12961 seconds, throughput 2095 KBytes/sec DUMP: Closing /dev/null DUMP: DUMP IS DONE For comparison: huff@>> dd if=/backup/Tue/usr/2007.Jan.30.usr.dump.tgz of=/dev/null 901411+1 records in 901411+1 records out 461522455 bytes transferred in 40.817512 secs (11306972 bytes/sec) > >huff@>>cat /tmp/null_dump > ... > > DUMP: DUMP: 27485576 tape blocks on 1 volume > > DUMP: finished in 13818 seconds, throughput 1989 KBytes/sec > > That is not good. Presumably the dmesg output looks sane (ie it's > not running narrow SCSI-1) and dd (or similar) report decent > thruput. kernel: ahc0: <Adaptec 2940 Ultra2 SCSI adapter> port 0xa400-0xa4ff mem 0xf1000000-0xf1000fff irq 11 at device 8.0 on pci0 kernel: ahc0: [GIANT-LOCKED] jerusalem kernel: da1 at ahc0 bus 0 target 4 lun 0 jerusalem kernel: da1: <SEAGATE SX150176LC BA11> Fixed Direct Access SCSI-2 device jerusalem kernel: da1: 80.000MB/s transfers (40.000MHz, offset 15, 16bit) jerusalem kernel: da1: Command Queueing Enabled jerusalem kernel: da1: 47702MB (97693755 512 byte sectors: 255H 63S/T 6081C) > Can you provide some more details on the FS? What are the block > and frag sizes It's UFS2, with default Block and frag sizes. > and how many inodes are used? huff@>>df -i | grep usr Filesystem 1024-blocks Used Avail Capacity iused ifree %iused Mounted on /dev/da1s1d 46287340 26883474 15700880 63% 423220 5582538 7% /usr Robert Huff
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?17855.41967.928998.142397>