Very Computer

   What is the best way to bring down two servers each running NFS?  I
have two SPARCServers 1000.  Each server mounts NFS filesystems from the
other server.  How do I bring both machines down at relatively the same
time.  Currently, I do a shutdown of one and then the other.  However,
the second server takes about a half-hour to come down because it must
wait to timeout when unmounting all the NFS-mounted filesystems which
reside on the other server.  I know there must be an easier way to do
this.  Any ideas / suggestions will be appreciated.  TIA

       Network Software Specialist     LMS Annex Room 105
       Kutztown University             phone: 610-683-4474
       Kutztown, PA. 19530             fax:   610-683-4634

  "You can bring a person to the university, but you can't make him think."

writes:

Quote:>>...  Each server mounts NFS filesystems from the other server.  How do I
>>bring both machines down at relatively the same time.  Currently, I do a
>>shutdown of one and then the other.  However, the second server takes about
>>a half-hour to come down because it must wait to timeout when unmounting all
>>the NFS-mounted filesystems which reside on the other server.

>The best way is *not* to mount filesystems from one server to the other, or
>automount them.  Alternatively you can unmount (by hand) the filesystems on
>the one before shutting down the other.

This has just reminded me of something I've always thought was wrong.  On the
way back up you get in trouble if you cross-mount because the system tries to
NFS mount filesystems off other machines before it exports its own, hence each
of two servers brought up at the same time will end up* on the other.
Surely this is the wrong way round ?  I always used to swap it on our 4.1.3
machines (export before import) before we trusted the automounter enough to
use it exclusively.  On Solaris 2, nfs.client and nfs.server are even done at
different run levels, so this seems a more fundamental step to take.

Cheers,

Rob
--
UUCP:   ...!mcsun!uknet!warwick!cudcv   PHONE:  +44 203 523037

Rob McMahon, Computing Services, Warwick University, Coventry CV4 7AL, England

If you add the bg option, the server will only hang a relatively shortwhile
before continueing.  The timeout is pretty quick, once the other machine
has portmapper started.  (RPC will return an error code because there's no
mountd registered on the otehr side yet).

But really: a situation like that calls for the automounter.

Casper

Quote:>   What is the best way to bring down two servers each running NFS?  I
>have two SPARCServers 1000.  Each server mounts NFS filesystems from the
>other server.  How do I bring both machines down at relatively the same
>time.  Currently, I do a shutdown of one and then the other.  However,
>the second server takes about a half-hour to come down because it must
>wait to timeout when unmounting all the NFS-mounted filesystems which
>reside on the other server.  I know there must be an easier way to do
>this.  Any ideas / suggestions will be appreciated.  TIA

The best way is *not* to mount filesystems from one server to the
other, or automount them.  Alternatively you can unmount (by hand)
the filesystems on the one before shutting down the other.

Or don't unmount the NFS filesystems at all.  (Remove the nfs entries
from /etc/mnttab prior to shutting down)

Why do you shut your servers down anyway?

Casper