From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from fieldses.org ([174.143.236.118]:53745 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752146Ab0IBSuU (ORCPT ); Thu, 2 Sep 2010 14:50:20 -0400 Date: Thu, 2 Sep 2010 14:49:43 -0400 From: "J. Bruce Fields" To: Steve Dickson Cc: Chuck Lever , Jeff Layton , Neil Brown , linux-nfs@vger.kernel.org Subject: Re: [PATCH] rpc.nfsd: mount up nfsdfs is it doesn't appear to be mounted yet (try #2) Message-ID: <20100902184943.GA24608@fieldses.org> References: <1283283160-30024-1-git-send-email-jlayton@redhat.com> <20100901204818.GA10507@fieldses.org> <20100901165605.41f4978d@tlielax.poochiereds.net> <20100902073156.577757b5@notabene> <4C7F8A80.3050705@RedHat.com> <20100902075518.58a1daf3@corrin.poochiereds.net> <8310F0E2-777C-4F17-B133-766FBAD30E19@oracle.com> <20100902142501.GD13117@fieldses.org> <4C7FD3D1.30601@RedHat.com> Content-Type: text/plain; charset=us-ascii In-Reply-To: <4C7FD3D1.30601@RedHat.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On Thu, Sep 02, 2010 at 12:41:53PM -0400, Steve Dickson wrote: > > > On 09/02/2010 10:25 AM, J. Bruce Fields wrote: > > On Thu, Sep 02, 2010 at 10:04:19AM -0400, Chuck Lever wrote: > >> Neil's patches are now in upstream nfs-utils. You can disable all of > >> the network listeners with command line options on mountd. > >> > >> It's been said that people complain that they don't like running > >> rpc.mountd on NFSv4-only servers. I'm not sure why that's a problem > >> we have to fix with a code change. Better documentation, better > >> automatic configuration detection in the NFS start-up scripts, or > >> simply renaming rpc.mountd could solve this issue without the need for > >> rip-and-replace of well-tested code. > I think a lot of the problems come from people having to open up > their firewalls and such... basically security issues.. was well > as having "extra" daemons stealing cpu cycles... > > > > > Yeah, I've suggested a separate upcall-handling daemon before, but agree > > that for now we should leave well enough alone. > What would be the deciding factor to start this work? Maybe when > Trond done splitting up kernel code into separate version modules? I think it's at most an idea to keep in the back of our minds in case it looks useful some day--not a todo we should schedule. --b.