From owner-freebsd-hackers  Thu Feb  2 00:23:01 1995
Return-Path: hackers-owner
Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id AAA21074 for hackers-outgoing; Thu, 2 Feb 1995 00:23:01 -0800
Received: from tree.com (tree.com [204.91.58.2]) by freefall.cdrom.com (8.6.9/8.6.6) with SMTP id AAA21068 for <freebsd-hackers@freebsd.org>; Thu, 2 Feb 1995 00:22:56 -0800
Received: by tree.com (5.0/SMI-SVR4)
	id AA05341; Thu, 2 Feb 95 03:21:34 EST
Date: Thu, 2 Feb 95 03:21:34 EST
From: ups@tree.com (Stephan Uphoff)
Message-Id: <9502020821.AA05341@tree.com>
To: freebsd-hackers@FreeBSD.org
Subject: Re: Am I dreaming?
Cc: ups@tree.com
X-Sun-Charset: US-ASCII
content-length: 1390
Sender: hackers-owner@FreeBSD.org
Precedence: bulk


>> amd relies on NFS retransmissions for reliability.  If it needs to do
>> something that isn't finished right away, it will just drop the
>> packet, start the operation, and let the kernel eventually time out

>I think the question was more one of how it get its mitts on user
>requests for files that aren't currently mounted.
>
>					Jordan


In order to automout nfs filesystems you need a special filesystem
that sits in the path of file that the user requests.
When a user requests a file it passes through this filesystem.
The filesystem can at that point mount the remote filesystem.
Confused ? Me too - sorry I'm still on a jetlag :(

Maybe an example will help:

@@@@@
A directory contains  auto-mountpoints

/auto

The use should be able to automount a:/home1 and B:/home2 as
/auto/home1 and /auto/home2.

The trick is:
	/auto is a special filesystem (a special nfs server)
        that is triggered by lookup request in the directory /auto
        for home2.
        It then mounts B:/home2 on /mnt_points/XXX (or whatever)
        and the lookup resturns a symbolic link to /mnt_points/XXX
        Normal NFS will now work with the remote filesystem
@@@@@        
 
I hope that answers the questions. 
There are actually two methods I know of - so if you are still
interested let me know and I'll try for a full explanation tomorrow
(when I'm not sleeptyping)


	Stephan