mountd keeps vanishing (!)

[eric () ee pdx edu (Eric Berggren)]

  we have a host that is creating many problems as it's mountd stops
responding to requests (and mucking up automounters all over). the
process is still running but "rpcinfo -u hostname 100005" results
in:

        rpcinfo: RPC: Timed out
        program 100005 version 0 is not available

  it's an SS2 running 4.1.3 with the following patches (among many others) :

    100173-10   NFS Jumbo Patch
P s 100249-09   automounter jumbo patch

  this would happen various hosts on rare occasions, but this box has now
done this 6 times within the last 2 days (it's a lab support box) with
anywhere from minutes to hours between occurences. the major problem is
that other hosts go ballistic during this time, including the central
fileservers.


  it's hard to figure out exactly what's happening as by the time it's
noticed, the machine itself is inaccessible to logins and many other
commands. we may try enabling full accounting (to whatever effect).

  anyone seen this or have any insight ? unfortunately, the automounter
is a critical part of our setup.

  thanx!

-eric

=============================================================================
-                         Portland State University                         -
Eric Berggren                           Janaka Jayawardena
Administrator, CS/EE                    Director of Computer Services, CS/EE
eric () ee pdx edu                         janaka () ee pdx edu