From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262406AbUDPQTD (ORCPT ); Fri, 16 Apr 2004 12:19:03 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263371AbUDPQTD (ORCPT ); Fri, 16 Apr 2004 12:19:03 -0400 Received: from adsl-207-214-87-84.dsl.snfc21.pacbell.net ([207.214.87.84]:10881 "EHLO lade.trondhjem.org") by vger.kernel.org with ESMTP id S262406AbUDPQTA (ORCPT ); Fri, 16 Apr 2004 12:19:00 -0400 Subject: Re: NFS export ... again From: Trond Myklebust To: fabian.frederick@prov-liege.be Cc: linux-kernel@vger.kernel.org In-Reply-To: References: Content-Type: text/plain Content-Transfer-Encoding: 7bit Message-Id: <1082132338.2581.42.camel@lade.trondhjem.org> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Fri, 16 Apr 2004 09:18:59 -0700 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, 2004-04-16 at 00:05, fabian.frederick@prov-liege.be wrote: > Trond, Chris, > > AFAICS, nfsd/export code already does some nfs cache poll through RPC to generate /proc/nfs/exports .... But content appears broken to me as it relates progressive mounting points but never umounts.We only have to fix this in my opinion ... or maybe I'm lured by multi-versioning there ... /proc/nfs/exports is a server-side thing, and just reflects the kernel's cache of currently exported filesystems. It is controlled by mountd by means of a syscall... I don't see what you are getting at here. There is no such thing as a reliable way of ensuring that the server has a fully uptodate list of all clients that are currently mounting a filesystem: Clients can and do crash. When they do, they will fail to send a "umount" call to the server. When that happens, the reference count in /var/lib/nfs/rmtab will be wrong, and so all these lists from /proc/fs/nfs/exports to "showmount" will show clients that are no longer there. Just don't rely on that information: it will never be completely correct. Cheers, Trond