From: Trond Myklebust Subject: Re: Insane permissions problem :) Date: Thu, 18 Jul 2002 14:25:31 +0200 Sender: nfs-admin@lists.sourceforge.net Message-ID: <15670.46011.421512.863931@charged.uio.no> References: <20020717225841.GC4853@netlocal.com.br> <15670.39323.156232.887895@notabene.cse.unsw.edu.au> Reply-To: trond.myklebust@fys.uio.no Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Joel Franco =?iso-8859-1?q?Guzm=E1n?= , nfs@lists.sourceforge.net Return-path: Received: from pat.uio.no ([129.240.130.16]) by usw-sf-list1.sourceforge.net with esmtp (Exim 3.31-VA-mm2 #1 (Debian)) id 17VALv-0006mA-00 for ; Thu, 18 Jul 2002 05:25:51 -0700 To: Neil Brown In-Reply-To: <15670.39323.156232.887895@notabene.cse.unsw.edu.au> Errors-To: nfs-admin@lists.sourceforge.net List-Help: List-Post: List-Subscribe: , List-Id: Discussion of NFS under Linux development, interoperability, and testing. List-Unsubscribe: , List-Archive: >>>>> " " == Neil Brown writes: > Trond: is there a good reason why the client cannot make sure > the groupid of a file is sorted first when accessing that file? > It would solve 99% of these problems in a very simple way and > should (says he who hasn't looked at the code) not be too hard > to achieve. It could be done, but doing so might break attempts to provide optimal reordering of groups by hand. The latter is needed in order to solve the remaining 1% of problems if, for instance, uid/gid mapping is being used on the server. I'm also reluctant to implement a solution that would involve giving the NFS code intimate knowledge of the details of the RPC auth handling. After all, the problem+solution you outline only effects AUTH_SYS (a.k.a. AUTH_UNIX). The workaround if one can't mess with /etc/groups in order to optimize the user's group membership at login is to use the 'newgrp' command. Note: I believe Frank van Marseveen once implemented a solution in which membership of > 16 groups would result in the RPC code reordering the groups and then retrying requests that fail with EACCES. I don't know if he still maintains that code, but he did claim that it worked for his case. Cheers, Trond ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs