Date: Wed, 7 Mar 2007 23:12:38 GMT From: Alan Amesbury<amesbury@umn.edu> To: freebsd-gnats-submit@FreeBSD.org Subject: docs/110061: [PATCH] tuning(7) missing reference to vfs.read_max Message-ID: <200703072312.l27NCcgd070976@www.freebsd.org> Resent-Message-ID: <200703072320.l27NK4QL048762@freefall.freebsd.org>
index | next in thread | raw e-mail
>Number: 110061 >Category: docs >Synopsis: [PATCH] tuning(7) missing reference to vfs.read_max >Confidential: no >Severity: non-critical >Priority: low >Responsible: freebsd-doc >State: open >Quarter: >Keywords: >Date-Required: >Class: update >Submitter-Id: current-users >Arrival-Date: Wed Mar 07 23:20:04 GMT 2007 >Closed-Date: >Last-Modified: >Originator: Alan Amesbury >Release: 6.1-RELEASE-p10 >Organization: University of Minnesota >Environment: FreeBSD paulaner.oitsec.umn.edu 6.1-RELEASE-p10 FreeBSD 6.1-RELEASE-p10 #5: Tue Oct 3 13:31:01 CDT 2006 root@paulaner.oitsec.umn.edu:/usr/obj/usr/src/sys/OITSEC-6-IS i386 >Description: The tuning(7) manpage is missing documentation for vfs.read_max. Since this knob can have a measurable, sometimes drastic, improvement on read performance for some systems, I think it deserves some mention. >How-To-Repeat: Read tuning(7) and note absence of mention of vfs.read_max. >Fix: This patch, derived against $FreeBSD: src/share/man/man7/tuning.7,v 1.72 2005/06/30 13:04:25 hmp Exp $ may work. ---------- Patch for tuning(7) ---------- --- tuning.7.ORIG Wed Mar 7 16:37:36 2007 +++ tuning.7 Wed Mar 7 17:04:37 2007 @@ -464,6 +464,13 @@ higher write queueing values may add latency to reads occurring at the same time. .Pp +The +.Va vfs.read_max +sysctl (default 8) determines how many blocks to read ahead during sequential +reads. For system loads where lots of sequential reading is done, +increasing this to 16 or 32 may increase performance, particularly for +RAID systems. +.Pp There are various other buffer-cache and VM page cache related sysctls. We do not recommend modifying these values. As of -------- End patch for tuning(7) -------- I wasn't able to attach the patch using the webform's "And/or patch file" option, as it didn't seem to think it was a patch. >Release-Note: >Audit-Trail: >Unformatted:help
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200703072312.l27NCcgd070976>