Date: Wed, 22 Aug 2007 18:59:02 +0800 From: Ganbold <ganbold@micom.mng.net> To: =?UTF-8?B?RGFnLUVybGluZyBTbcO4cmdyYXY=?= <des@des.no> Cc: questions@freebsd.org, freebsd-amd64@freebsd.org Subject: Re: What is a "sane" setting for maxdsize when running amd64? it seems many normal suggestions do not apply. Message-ID: <46CC16F6.7020904@micom.mng.net> In-Reply-To: <86r6lvalht.fsf@ds4.des.no> References: <835936.35104.qm@web34510.mail.mud.yahoo.com> <A502D00B-80A7-4BC6-9842-D0A2A50E2026@mac.com> <86r6lvalht.fsf@ds4.des.no>
next in thread | previous in thread | raw e-mail | index | archive | help
Dag-Erling Smørgrav wrote: > Chuck Swiger <cswiger@mac.com> writes: > >> You should configure squid to use no more than about 60 - 70% of the >> available physical RAM-- ie, set the cache_mem parameter to about 2.5 >> or 3GB. >> > > Better yet, don't run Squid at all. Ok, then what do you recommend instead of Squid? thanks, Ganbold > It was designed for a computer > architecture that was already obsolete when Squid was first written. > > >> It wouldn't be unreasonable to limit datasize to 3 GB on such a >> machine, assuming that nothing you run will ever need to grow >> larger... >> > > ..actually, maxdsiz is meaningless in FreeBSD 7, because the new > allocator uses mmap(2) instead of brk(2) / sbrk(2), so malloc() counts > towards the resident set size (ulimit -m), not the data segment size > (ulimit -d). > > (unless, of course, your application has its own allocator, in which > case you can kiss performance goodbye) > > DES > -- Heuristics are bug ridden by definition. If they didn't have bugs, then they'd be algorithms.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?46CC16F6.7020904>