From mboxrd@z Thu Jan 1 00:00:00 1970 From: "J. Bruce Fields" Subject: Re: processes in D state too long too often Date: Tue, 10 Feb 2009 11:02:58 -0500 Message-ID: <20090210160258.GD29075@fieldses.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-kernel@vger.kernel.org, linux-nfs@vger.kernel.org, Andrew Morton To: "Gary L. Grobe" Return-path: Received: from mail.fieldses.org ([141.211.133.115]:40655 "EHLO pickle.fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754583AbZBJQCv (ORCPT ); Tue, 10 Feb 2009 11:02:51 -0500 In-Reply-To: Sender: linux-nfs-owner@vger.kernel.org List-ID: On Tue, Feb 10, 2009 at 02:18:33AM +0000, Gary L. Grobe wrote: > > >>The more likely explanation is that you just switched to a more recent > >>distro where "sync" (as opposed to "async") is the option. Depending on > >>workload, "async" may improve performance a great deal, at the expense > >>of possible data corruption on server reboot! > >> > >>If you're doing a lot of writing and using NFSv2, then switching to > >>NFSv3 may give you performance close to the "async" performance without > >>the corruption worries. > > Just a small update about our rollback I need to correct. Turns out our problem has been solved by going with the 2.6.20-r10 of the gentoo-sources patched kernel. Although gentoo marks this as unstable for amd64, it's working fine. I've made no other changes than going back a few versions on the kernel and adjusting the .config w/ the same settings. > > Tomorrow I'll likely give the next marked stable patched gentoo-sources kernel another try which was 2.6.24-r10 and recheck my configs and try to gather anything else I can gather from it. Sounds good, thanks. --b.