From owner-freebsd-stable@FreeBSD.ORG Fri Nov 30 07:17:05 2007 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 2CD7016A476 for ; Fri, 30 Nov 2007 07:17:05 +0000 (UTC) (envelope-from lol@chistydom.ru) Received: from hermes.hw.ru (hermes.hw.ru [80.68.240.91]) by mx1.freebsd.org (Postfix) with ESMTP id 8EF9413C469 for ; Fri, 30 Nov 2007 07:17:03 +0000 (UTC) (envelope-from lol@chistydom.ru) Received: from [80.68.244.40] (account a_popov@rbc.ru [80.68.244.40] verified) by hermes.hw.ru (CommuniGate Pro SMTP 5.0.14) with ESMTPA id 204260245; Fri, 30 Nov 2007 10:16:40 +0300 Message-ID: <474FB8CA.8020608@chistydom.ru> Date: Fri, 30 Nov 2007 10:16:26 +0300 From: Alexey Popov User-Agent: Thunderbird 2.0.0.6 (X11/20070924) MIME-Version: 1.0 To: Pete French References: In-Reply-To: Content-Type: text/plain; charset=KOI8-R; format=flowed Content-Transfer-Encoding: 7bit Cc: killing@multiplay.co.uk, freebsd-stable@freebsd.org, max@love2party.net Subject: Re: Also seeing 2 x quad-core system slower that 2 x dual core X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 30 Nov 2007 07:17:05 -0000 Hi Pete French wrote: >> Have you checked that your dir hash isn't suffering due to lack of memory >> this can have a marked impact on seemingly trivial things like this as >> could silly things like the RAID card being installed in a different slot. > > RAID card is onboard on these things - how would I check the dir hash ? The > slower server has 16 gig of RAM, the faster one has 4 gig. Both were > installed the same way in the same order, so should have the same disc layout > more or less. dirhash is a cache of directories that store many files. FreeBSD has unreasonably small default for the maximum size of this cache. See sysctl -a | grep dirhash and tune up vfs.ufs.dirhash_maxmem if vfs.ufs.dirhash_mem is near to the limit. Also could you please provide `top` and `systat -vm 1` output on dual- and quad-core systems during your tests? With best regards, Alexey Popov