Date: Wed, 31 Aug 2011 12:19:20 +0400 From: Lev Serebryakov <lev@FreeBSD.org> To: Daniel Kalchev <daniel@digsys.bg> Cc: freebsd-fs@freebsd.org Subject: Re: Very inconsistent (read) speed on UFS2 Message-ID: <10310173523.20110831121920@serebryakov.spb.ru> In-Reply-To: <4E5DDE66.5000508@digsys.bg> References: <1945418039.20110830231024@serebryakov.spb.ru> <317753422.20110830231815@serebryakov.spb.ru> <20110831004251.GA89979@icarus.home.lan> <4E5DDE66.5000508@digsys.bg>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello, Daniel. You wrote 31 =E0=E2=E3=F3=F1=F2=E0 2011 =E3., 11:10:30: > This is very probable, given RAID5, that needs to read stripes off every > disc on every read. Again: how faulty drive could give inconsistency in one file? If it has relocated sectors (additional long seek in some place), it will give consistent performance degradation in that place. > Probably some S.M.A.R.T. investigation might help (different=20 > measurements for different drives, if supported by the drives, that is). SMARTs are almost identical and exclellent. No relocated sectors (at all!) not multizone read errors (at all!), etc. I'll try to synchronize iostat statistic and reading benchmark, but it is hard to do, as I can not think of the way how to know, that these 100 lines of iostat output is exactly this (slow) reading of this file, when there is 10+ files in question and all of them, are read 10+ times. --=20 // Black Lion AKA Lev Serebryakov <lev@FreeBSD.org>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?10310173523.20110831121920>