From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Pearson Subject: Re: [bug] FC2/2.6.8.1 does not fallback to NFSv2 Date: Sat, 28 Aug 2004 17:47:47 +0100 Sender: nfs-admin@lists.sourceforge.net Message-ID: <4130B733.7070506@moving-picture.com> References: <20040827212244.GA57703@brandon.dvalentine.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Cc: nfs@lists.sourceforge.net 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 1C16Me-0002Jv-72 for nfs@lists.sourceforge.net; Sat, 28 Aug 2004 09:47:40 -0700 Received: from mpc-26.sohonet.co.uk ([193.203.82.251] helo=moving-picture.com) by sc8-sf-mx1.sourceforge.net with esmtp (Exim 4.34) id 1C16Mc-0006Fp-G2 for nfs@lists.sourceforge.net; Sat, 28 Aug 2004 09:47:40 -0700 To: "Brandon D. Valentine" In-Reply-To: <20040827212244.GA57703@brandon.dvalentine.com> 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: It's a bug in Redhat's util-linux package - see: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=123416 As yet, I haven't found a fixed version available from Redhat - my simple workround is as follows - I guess it's OK, as this check doesn't seem to exist in other versions of nfsmount.c James Pearson *** util-linux-2.12/mount/nfsmount.c.dist 2004-07-08 15:08:38.000000000 +0 100 --- util-linux-2.12/mount/nfsmount.c 2004-07-08 15:15:28.000000000 +0100 *************** *** 423,430 **** --- 423,435 ---- return 1; memcpy(mnt_pmap, &save_mnt, sizeof(*mnt_pmap)); } + + /* the following check breaks the automatic fall back + to NFSv2 - JCP 8/7/2004 */ + #if 0 if (rpc_createerr.cf_stat != RPC_PROGNOTREGISTERED) break; + #endif memcpy(nfs_pmap, &save_nfs, sizeof(*nfs_pmap)); } out_bad: Brandon D. Valentine wrote: > [ I tried to send this yesterday without actually subscribing, but it > has yet to make it past the list moderator. I've given up and > subscribed to yet another mailing list to get this bug report out. ] > > We have just brought up several Fedora Core 2 systems running the latest > 2.6.8.1 kernel revision in our NIS/NFS+autofs environment. The FC2/2.6 > clients are refusing to mount one specific filesystem on our network. > This filesystem is exported by a Debian/woody system running a 2.4.18 > kernel. The kernel on this machine only supports NFSv2 exports. > > When the FC2/2.6 client attempts to automount the filesystem on the 2.4 > server we observe the following in the client's syslog: > > Aug 26 18:37:08 scylla automount[5056]: >> mount to NFS server 'reef' > failed: server is down. > Aug 26 18:37:08 scylla automount[5056]: mount(nfs): nfs: mount failure > reef:/export/reef/httpd/ on /sb/www > Aug 26 18:37:08 scylla automount[5056]: failed to mount /sb/www > > And the following is observed in the server's syslog: > > Aug 26 18:37:12 reef kernel: svc: unknown version (3) > > At this point the client gives up with the following error: > > [root@scylla root]# ls /sb/www > ls: /sb/www: No such file or directory > > None of our other Linux systems running 2.4 kernels have any trouble > mounting that filesystem. They all automount it using NFSv2 as > expected. All of our 2.4 systems also correctly mount using NFSv3 from > systems that support it. > > Autofs is configured on the FC2 systems to permit NFSv2 mounts as shown > by the minproto option below: > > [root@scylla root]# mount | grep automount > automount(pid1665) on /home type autofs (rw,fd=4,pgrp=1665,minproto=2,maxproto=4) > automount(pid1695) on /sb type autofs (rw,fd=4,pgrp=1695,minproto=2,maxproto=4) > automount(pid1739) on /scratch type autofs (rw,fd=4,pgrp=1739,minproto=2,maxproto=4) > > The 2.6.8.1 kernel is configured with the following options: > > [root@scylla root]# grep -i nfs /usr/src/linux/.config > CONFIG_NFS_FS=y > CONFIG_NFS_V3=y > # CONFIG_NFS_V4 is not set > # CONFIG_NFS_DIRECTIO is not set > CONFIG_NFSD=y > CONFIG_NFSD_V3=y > # CONFIG_NFSD_V4 is not set > CONFIG_NFSD_TCP=y > > I am able to manually mount this filesystem, but only if I explicitly > add the vers=2 option to the mount command: > > [root@scylla root]# mount -t nfs reef:/export/reef/httpd/ /mnt/temp > mount to NFS server 'reef' failed: server is down. > [root@scylla root]# mount -t nfs -o vers=2 reef:/export/reef/httpd/ /mnt/temp > [root@scylla root]# > > Given the fact that the mount fails manually until vers=2 is specified I > am inclined to believe this is a kernel NFS bug, not an autofs or mount > bug. > > I would be glad to gather any additional information that might be > necessary to determine the cause of this behavior. Can anyone reproduce > this or have advice on how I might fix it? > > Thanks, > > Brandon D. Valentine ------------------------------------------------------- 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