From owner-freebsd-hackers Tue Jul 18 15:16:06 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id PAA18514 for hackers-outgoing; Tue, 18 Jul 1995 15:16:06 -0700 Received: from devnull (devnull.mpd.tandem.com [131.124.4.29]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id PAA18497 ; Tue, 18 Jul 1995 15:16:02 -0700 Received: from olympus by devnull (8.6.8/8.6.6) id RAA19269; Tue, 18 Jul 1995 17:15:53 -0500 Received: by olympus (4.1/TSS2.1) id AA29792; Tue, 18 Jul 95 17:15:55 CDT From: faulkner@mpd.tandem.com (Boyd Faulkner) Message-Id: <9507182215.AA29792@olympus> Subject: Re: FBSD v2.0.5: NFS to multi-homed server broken? To: mikebo@tellabs.com (Mike Borowiec) Date: Tue, 18 Jul 1995 17:15:54 -0500 (CDT) Cc: questions@freebsd.org, hackers@freebsd.org In-Reply-To: from "Mike Borowiec" at Jul 18, 95 05:01:13 pm X-Mailer: ELM [version 2.4 PL17] Content-Type: text Content-Length: 1729 Sender: hackers-owner@freebsd.org Precedence: bulk > > faulkner@mpd.tandem.com (Boyd Faulkner) wrote: > > mikebo@tellabs.com (Mike Borowiec) wrote: > > > I have found some anomolous NFS behaviour in FreeBSD 2.0.5R. > > > > > > Problem 1: > > > FreeBSD attempts to NFS mount a partition from a multi-homed server not > > > on the same ethernet, and the return route from the server to the FreeBSD > > > client is via another interface (with a different hostname and IP address). > > > If done manually, the mount command can be killed. If the mount is in the > > > /etc/fstab file, the machine hangs on startup and must be reset. > > > > Been there, done that. You need the noconn option. This way, when it comes > > back along a different route, there is not a connnection already in force. > > I think this is the -c option if you use mount_nfs directly. Anyway, > > -o noconn solved the problem for me. > > > This may work if the portmapper reply port number is non-standard > ( != 111/udp ), but having tried numerous combinations of conn, noconn > and mount_nfs -c, it appears to me that it does nothing when the > portmapper reply is from the proper host, but from a different IP address. > > Does anyone else have any clue how this works in SunOS? > - Mike > -- Well, it worked for me in 2.0, for exactly the problem you've cited. But this WAS under the guise of amd, the automount daemon. Now I am getting NFS Portmap: RPC: Port mapper failure - RPC: Timed out. If I get this resolved, I'll let you know. amd still works though. Boyd -- _______________________________________________________________________ Boyd Faulkner - faulkner@isd.tandem.com - http://cactus.org/~faulkner _______________________________________________________________________