From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932368AbWEXQWA (ORCPT ); Wed, 24 May 2006 12:22:00 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S932378AbWEXQWA (ORCPT ); Wed, 24 May 2006 12:22:00 -0400 Received: from pat.uio.no ([129.240.10.4]:56489 "EHLO pat.uio.no") by vger.kernel.org with ESMTP id S932368AbWEXQV7 (ORCPT ); Wed, 24 May 2006 12:21:59 -0400 Subject: Re: Linux 2.6 NFS client read-ahead From: Trond Myklebust To: German San Agustin Cc: linux-kernel@vger.kernel.org In-Reply-To: <2bc0baf20605240453l89e5cd7w949377ead93e8b66@mail.gmail.com> References: <2bc0baf20605240446h197b3d2fxc797404aa0e733ba@mail.gmail.com> <2bc0baf20605240453l89e5cd7w949377ead93e8b66@mail.gmail.com> Content-Type: text/plain Date: Wed, 24 May 2006 12:21:45 -0400 Message-Id: <1148487705.5872.34.camel@lade.trondhjem.org> Mime-Version: 1.0 X-Mailer: Evolution 2.6.1 Content-Transfer-Encoding: 7bit X-UiO-Spam-info: not spam, SpamAssassin (score=-3.826, required 12, autolearn=disabled, AWL 1.17, UIO_MAIL_IS_INTERNAL -5.00) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 2006-05-24 at 13:53 +0200, German San Agustin wrote: > We have several linux 2.6.9 accesing to a netapp filer via nfs to > update several thousands of files. The application only need to read a > few blocks of the files when updating so we found out that disabling > the read-ahead on the server improve greatly the performance by > maintaining a clean cache and decreasing the number of access to disk. > We have been trying to disable the read-ahead in the client as well to > reduce the access to the server even further; but we couldn't find > where to do this in the 2.6 kernel family. Is it possible?, or it is > simply that on 2.6 kernels it is not possible to tune the nfs client. It is not possible, and in this case, it really shouldn't make much difference to you either as the readahead code is adaptive: it should automatically detect that you are doing short reads. If you want to check that it is working, I suggest using tcpdump to monitor the NFS traffic to that file, then comparing with a "strace" dump. Cheers, Trond