From: Trond Myklebust Subject: Re: userspace NFS daemon Date: Sun, 05 Sep 2004 23:32:57 -0400 Sender: nfs-admin@lists.sourceforge.net Message-ID: <1094441577.10492.134.camel@lade.trondhjem.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Cc: Frank van Maarseveen , Linux NFS mailing list Return-path: Received: from sc8-sf-mx1-b.sourceforge.net ([10.3.1.11] helo=sc8-sf-mx1.sourceforge.net) by sc8-sf-list2.sourceforge.net with esmtp (Exim 4.30) id 1C4AFp-00064K-FV for nfs@lists.sourceforge.net; Sun, 05 Sep 2004 20:33:17 -0700 Received: from pat.uio.no ([129.240.130.16] ident=7411) by sc8-sf-mx1.sourceforge.net with esmtp (Exim 4.34) id 1C4AFm-0005VW-5j for nfs@lists.sourceforge.net; Sun, 05 Sep 2004 20:33:17 -0700 To: Bogdan Costescu In-Reply-To: Errors-To: nfs-admin@lists.sourceforge.net List-Unsubscribe: , List-Id: Discussion of NFS under Linux development, interoperability, and testing. List-Post: List-Help: List-Subscribe: , List-Archive: P=E5 su , 05/09/2004 klokka 20:47, skreiv Bogdan Costescu: > Err, you mean to say that root-NFS only appeared after knfsd ? And=20 > that knfsd was absolutely flawless since day 1 ? :-) No. I mean that very few people have ever used unfsd in mission-critical production environments. You might get away with it for read-only stuff (if you are prepared to put up with the speed), but nobody should be using it for read/write partitions. > Cluster nodes have usually a small number of software packages=20 > installed. Most "core" packages (like core-utils, grep, gawk, etc.)=20 > don't need locking. As / doesn't see itself much writting (which=20 > happens usually in /var and /tmp, but those can be mounted=20 > separately), NFSv3 vs. NFSv2 doesn't make much difference either. So=20 > what's the problem with using unfsd in these conditions ? If you have to mount all the writeable partitions separately from a more robust server, then why bother? Replicating /etc and changing the two or three files that differ for each cluster node is hardly a difficult task: you can do it using a simple shell script. /bin doesn't even have to be replicated... The "per-uid" and "per-gid" stuff might be more useful, but is likely to be defeated by lookup caching rules on most NFS clients. Note: the premise: If a matching file is located, but authorization is denied,this is NOT considered a match, and the next entry on the list will be attempted. if taken literally, would certainly make managing these name.ipaddress files "interesting": "chmod a+r,a-w blah; cat >blah" might have some curious side-effects... Cheers, Trond ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs