Return-Path: Received: from mx2.netapp.com ([216.240.18.37]:37087 "EHLO mx2.netapp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752799Ab1C3JAt convert rfc822-to-8bit (ORCPT ); Wed, 30 Mar 2011 05:00:49 -0400 Subject: Re: NFS v4 client blocks when using UDP protocol From: Trond Myklebust To: Tom Haynes Cc: peter.staubach@emc.com, Zoltan.Menyhart@bull.net, chuck.lever@oracle.com, linux-nfs@vger.kernel.org In-Reply-To: <20110330001232.GB19700@adept.internal.excfb.com> References: <4D91ED79.4020302@bull.net> <4A05F533-1BE6-4DFC-BC53-9392DA8BED76@oracle.com> <4D91FA1D.1000206@bull.net> <5E6794FC7B8FCA41A704019BE3C70E8B069D40C4@MX31A.corp.emc.com> <20110330001232.GB19700@adept.internal.excfb.com> Content-Type: text/plain; charset="UTF-8" Date: Wed, 30 Mar 2011 11:00:47 +0200 Message-ID: <1301475647.7654.34.camel@lade.trondhjem.org> Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On Tue, 2011-03-29 at 19:12 -0500, Tom Haynes wrote: > On Tue, Mar 29, 2011 at 11:50:36AM -0400, peter.staubach@emc.com wrote: > > It does talk about NFS over UDP, interestingly, but the text on page 25, indicates that the transport of choice MUST be one of the IETF-approved congestion control transport protocols, of which, UDP is not one. > > > > Perhaps some clean up of RFC3530bis and RFC5661 could include removal of the UDP mentions. > > The text in 3530bis could be made to match that in 5661: > > It is permissible for a connectionless transport to be used under > NFSv4.1; however, reliable and in-order delivery of data combined > with congestion control by the connectionless transport is REQUIRED. > As a consequence, UDP by itself MUST NOT be used as an NFSv4.1 > transport. NFSv4.1 assumes that a client transport address and > server transport address used to send data over a transport together > constitute a connection, even if the underlying transport eschews the > concept of a connection. > > But as we can see 5661 is very strong. I think Peter is referring to the text in section 2.2 and section 17.2 which describes the format for netids and universal addresses specifically for the UDP case. I agree with Peter's suggestion that we should just delete that text. Cheers Trond -- Trond Myklebust Linux NFS client maintainer NetApp Trond.Myklebust@netapp.com www.netapp.com