[Mimedefang] Calling all Tru64 gurus (getting MD to work with Tru64)

[Jim McCullars]

Hi, I'm trying to get MD to work with HP Tru64 Unix (a/k/a Digital Unix
a/k/a OSF1) version 4.0F.  Here is what I have done so far and what still
doesn't work:

The default version of make does not like the trailing blank line in the
DEFS statement right above the "all: mimedefang mimedefang-multiplexor"
line.  Edit the generated Makefile to remove the escape \ on the last real
line and the tab-only line following it (or use GNU make).

When compiling, I got some wierd errors on
/usr/local/lib/sendmail/timers.h (said it was missing a semicolon).
David told me that MD shouldn't need that file so I renamed it.

When running the make, I would get undefined symbols relating to threads.
Following the suggestion of the pthread man page under Tru64, I added
"-pthread" to the end of all compile statements, and "-lpthread -lexc" to
the end of the link commands for mimedefang and mimedefang-multiplexor.

Undefined symbol "snprintf":  add -ldb to link statements.

Here's a wierd one - when I install perl modules as root, the directories
are left as no permissions for group or world.  That may be a umask
setting on our system, but it did bite me.

Now is the part where I am stuck.  When I start the multiplexor, it will
start, and start one slave.  When it tries to start a second slave, this
shows up in the syslog file:

Sep  9 14:10:27 email mimedefang-multiplexor: Starting slave 0 (pid 28696)
(1 running): Bringing slaves up to minSlaves
Sep  9 14:10:27 email mimedefang-multiplexor: Reap: Idle slave 0 (pid
28696) exited normally with status 1 (SLAVE DIED UNEXPECTEDLY)

It does this repeatedly until I stop the multiplexor.  The only other clue
I have is on the console, for each time it tries to start another slave,
this error gets logged:

28696: /usr/bin/perl: /sbin/loader: Fatal Error: cannot map libiconv.so

We have verified that libiconv.so is in a valid library path (we even
symlinked it to /usr/lib to make sure).

My gut feeling is that it has something to do with threads, since it seems
to start one slave process OK.  I just don't know where to turn next.  Any
ideas from this group?

Jim
*-------------------------------------------------------------------------*
* James H. McCullars                 I Phone: (256) 824-2610              *
* Director of Systems & Operations   I Fax:   (256) 824-6643              *
* Computer & Network Services        I Internet: mccullj at email.uah.edu    *
* The University of Alabama          I -----------------------------------*
*    in Huntsville                   I                                    *
* Huntsville, AL 35899               I This space for rent - CHEAP!       *
*-------------------------------------------------------------------------*