From: Chuck Lever Subject: Re: [PATCH 3/3] sunrpc: reduce timeout when unregistering rpcbind registrations. Date: Mon, 6 Jul 2009 14:32:41 -0400 Message-ID: <9D0F8CF2-3FA2-4065-A125-DE02C2674129@oracle.com> References: <20090528062730.15937.70579.stgit@notabene.brown> <20090528063303.15937.62423.stgit@notabene.brown> <18992.35996.986951.556723@notabene.brown> <4A51F125.5080709@suse.de> <4A52217E.9050207@suse.de> <4E8F91E6-4E55-44BB-889B-DDB9910129BF@oracle.com> <1246898450.11267.12.camel@heimdal.trondhjem.org> <68129579-E484-4E7E-B38D-4E14ED5A5B1D@oracle.com> <1246900456.11267.34.camel@heimdal.trondhjem.org> <71D4E90D-471B-4BFA-B47C-6A5BFD0754E9@oracle.com> <1246903105.23966.4.camel@heimdal.trondhjem.org> Mime-Version: 1.0 (Apple Message framework v935.3) Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Cc: Suresh Jayaraman , Neil Brown , Linux NFS mailing list To: Trond Myklebust Return-path: Received: from acsinet11.oracle.com ([141.146.126.233]:56703 "EHLO acsinet11.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751894AbZGFScs (ORCPT ); Mon, 6 Jul 2009 14:32:48 -0400 In-Reply-To: <1246903105.23966.4.camel-rJ7iovZKK19ZJLDQqaL3InhyD016LWXt@public.gmane.org> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Jul 6, 2009, at 1:58 PM, Trond Myklebust wrote: > On Mon, 2009-07-06 at 13:51 -0400, Chuck Lever wrote: >> In both cases, it turns out that the library implementations in user >> space already fail quickly. RPC_CANTRECV is returned if an attempt >> is >> made to send an rpcbind query to an inactive UDP port. >> RPC_SYSTEMERROR/ECONNREFUSED is returned if an attempt is made to >> send >> an rpcbind query to an inactive TCP port. In my view, the kernel is >> lacking here, and should be made to emulate user space more closely. > > I fully agree that we can fix the above by more judicious > interpretation > of the returned networking errors. What I don't see eye to eye with is > the assertion that has been floating about that the current behaviour > should be a good enough reason to add kernel AF_LOCAL support. Since portmapper doesn't support it, and the kernel (for now) needs backwards compatibility with portmapper, AF_LOCAL for rpcbind is rather a non-starter. In my opinion there are other reasons to consider kernel-level AF_LOCAL support at some later point. -- Chuck Lever chuck[dot]lever[at]oracle[dot]com