From owner-freebsd-fs@FreeBSD.ORG Fri May 14 08:21:22 2010 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 4F395106566B for ; Fri, 14 May 2010 08:21:22 +0000 (UTC) (envelope-from shurik@zk.informjust.ua) Received: from lambent.informjust.ua (lambent.informjust.ua [193.111.173.22]) by mx1.freebsd.org (Postfix) with ESMTP id DA8288FC08 for ; Fri, 14 May 2010 08:21:21 +0000 (UTC) Received: from status.informjust.ua ([10.1.10.202]) by lambent.informjust.ua with esmtp (Exim 4.71) (envelope-from ) id L2EHSL-000FN8-RN for freebsd-fs@freebsd.org; Fri, 14 May 2010 11:19:35 +0300 Received: from [192.168.72.1] (helo=zk.informjust.ua) by status.informjust.ua with esmtp (Exim 4.69 (FreeBSD)) (envelope-from ) id 1OCq8k-000Kne-PR for freebsd-fs@freebsd.org; Fri, 14 May 2010 11:21:02 +0300 Received: from monstro.zk.informjust.ua ([10.2.113.96]) by zk.informjust.ua with esmtp (Exim 4.68 (FreeBSD)) (envelope-from ) id 1OCq8K-0009r8-26 for freebsd-fs@freebsd.org; Fri, 14 May 2010 11:20:36 +0300 Message-ID: <4BED0810.10508@zk.informjust.ua> Date: Fri, 14 May 2010 11:21:36 +0300 From: "Alexander V. Ribchansky" User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; ru; rv:1.9.1.5) Gecko/20091217 Thunderbird/3.0 MIME-Version: 1.0 CC: freebsd-fs@freebsd.org References: <4BE95F1F.5090009@zk.informjust.ua> <20100511223813.GB3044@garage.freebsd.pl> In-Reply-To: <20100511223813.GB3044@garage.freebsd.pl> Content-Type: text/plain; charset=KOI8-R; format=flowed Content-Transfer-Encoding: 8bit X-Spam-Report: Spam detection software, running on the system "lambent.informjust.ua", has identified this incoming email as possible spam. The original message has been attached to this so you can view it (if it isn't spam) or label similar future email. If you have any questions, see root@localhost for details. Content preview: 12.05.2010 01:38, Pawel Jakub Dawidek пишет: > -skip- > Could you try the following patch without disabling UMA? > > http://people.freebsd.org/~pjd/patches/vfs_subr.c.7.patch > > It works quite well for me. > > Also we still don't have back-pressure mechanism when arc_meta_limit is > exceeded. OpenSolaris frees namecache entires then and we currently do > nothing. I'll experiment with a bit and let you know if the patch above > doesn't fix your problem still. > OOps, I do it again! ;) So after several hard days my STABLE gone away again. With your patch and UMA enabled. Wired constantly grow, last time I see it, it was about 940M As Wired grom, perfomance become terrible, buildworld take about 4 hours on Core2quadro.. So my wery IMHO (please don't kill me for it) that UMA allocation on i386 is unusable, correct me if I'm wrong. [...] Content analysis details: (-3.5 points, 5.0 required) pts rule name description ---- ---------------------- -------------------------------------------------- -1.8 ALL_TRUSTED Passed through trusted hosts only via SMTP 1.3 MISSING_HEADERS Missing To: header -2.6 BAYES_00 BODY: Bayesian spam probability is 0 to 1% [score: 0.0000] -0.4 AWL AWL: From: address is in the auto white-list X-Spam-Score: -3.5 (---) Subject: Re: Freebsd 8.0 kmem map too small X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 14 May 2010 08:21:22 -0000 12.05.2010 01:38, Pawel Jakub Dawidek пишет: > -skip- > Could you try the following patch without disabling UMA? > > http://people.freebsd.org/~pjd/patches/vfs_subr.c.7.patch > > It works quite well for me. > > Also we still don't have back-pressure mechanism when arc_meta_limit is > exceeded. OpenSolaris frees namecache entires then and we currently do > nothing. I'll experiment with a bit and let you know if the patch above > doesn't fix your problem still. > OOps, I do it again! ;) So after several hard days my STABLE gone away again. With your patch and UMA enabled. Wired constantly grow, last time I see it, it was about 940M As Wired grom, perfomance become terrible, buildworld take about 4 hours on Core2quadro.. So my wery IMHO (please don't kill me for it) that UMA allocation on i386 is unusable, correct me if I'm wrong. -- AVR39-RIPE