[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

trouble with nfs to remotely mounted Unix file systems



    Date: Fri, 9 Aug 1991 15:03 EDT
    From: rshapiro@arris.com (Richard Shapiro)

    >[context -- accessing host1:/foo/bar/bazz, which is a symbolic link
    > to host1:/net/host2/foo/bar/bazz, which is a local mounting of
    > foreign file system host2:/foo/bar/bazz]

    >Can you access HOST2:/foo/bar/bazz from the Lispm?  

    Yes. 

OK, that means that HOST2's exports are correct.

	 I don't want to do this because our sysadm wants the freedom to
    move file systems around whenever he needs to. I would prefer to do
    all Unix file system access through a single host (host1), and leave
    any foreign mounting etc up to that host.

That's a reasonable policy -- I try to do the same thing here, although
we mostly use automount maps instead of symbolic links so that we don't
have to duplicate the links on all the Unix systems.

    >Are you running Genera 8.0 or 8.1?  

    8.0.2 (UX1200) -- if there's an 8.1, we haven't received any word of
    it. 

It's been out since the beginning of June.

    >Do you have the NFS-AUTOMOUNT user properties set up properly in HOST1's
    >namespace entry?  

    I think so. I have it as: 

	NFS-AUTOMOUNT /net -hosts

    The Unix host in question invokes automounting via

	automount /net -hosts &&

That sure looks right to me.

What function is the error occurring in?  Is it happening in
NFS::MNT1PROC-MNT?  What host name is mentioned in the transport agent
argument, HOST1 or HOST2.

I'm grasping at straws, but could there be a discrepancy between the
address of HOST2 in the namespace and in HOST1's /etc/hosts or the
hosts.byname NIS map?  The Lispm tries to interpret /net/<host> the way
HOST1 would, so it looks in HOST1's hosts file and NIS map before
looking in the namespace.  If there are two different hosts being
referred to as HOST2, and the one in the namespace has the correct
exports but the one in the Unix hosts database doesn't, it could explain
what you're seeing.

                                                barmar