To: discuss@opennms.org
From: Pete Siemsen <siemsen@ucar.edu>
Date: Sun, 11 Mar 2001 21:04:35 -0700
Subject: [opennms-discuss] a visit to OpenNMS

I just spent a day and a half visiting the OpenNMS folks at their site
in North Carolina.  I thought I'd share some of what I learned, since
some of it is stuff I didn't know even after faithfully lurking on
this list.

First off, it was a pleasure to meet the team.  There are about 10
people in the group, and I was impressed with their experience,
knowledge and general brainpower.  This project is in good hands.  As
I assumed, it's a den of code development, with a casual environment
and a free-access refrigerator to make any serious programmer happy.

Before my visit, I had failed to install OpenNMS version 0.6 on my
laptop (750Mhz, 256M, SuSE 7.0, 2.2.17 kernel).  I brought the laptop
along, and several hours of hands-on help produced a fully-functional
version 7.0 system.  I know I wouldn't have been able to get it
running by myself.

Along the way I learned some things:

1. Installation is a bear.  They know this and are seriously hacking
   away at the problem.

2. If you've built your own kernel, as I had, you'll probably have to
   rebuild it to increase the maximum allowed tasks.  The default when
   you build a kernel is 256, and OpenNMS uses more than 400 threads.
   Edit your include/linux/tasks.h file and rebuild.  Users who run
   with the kernel that came with a popular distribution probably
   won't have a problem, as the distros are compiled with the value
   set high.

3. With the software up, a "ps" command is a little frightening, as
   it looks like there are 400+ processes on your machine.  Don't worry,
   it's an artifact of Java's thread support, and your system will run
   fine.

4. For now, you have to run the software as root.

5. PostgreSQL, Tomcat and icmpd need to be running before you start
   OpenNMS.

6. It takes, like, 2 minutes to start or stop the software.  For some
   configuration changes, you have to stop/start the software.  This
   will get better in the future.

7. Put the distribution in your home directory, in a subdirectory
   named opennms-all.  This is what the developers do.

8. You can use a JDK from Sun or from IBM.  They use IBM, but don't
   know of a reason not to use Sun's.  I have Sun's, so I'll find out.

9. There are two interfaces, the older "gui" or "Swing" interface and
   the quite new web interface.  For now, you have to run the gui
   interface for configuration tasks like setting up users or
   defining the limits of IP discovery.  Most of the time, you'll use
   the web interface to actually monitor your network.

10. Everything is configured by XML files.  The gui interface provides
   a friendly interface, but it helps if you're comfortable with XML.
   Unix types may prefer to edit the files.

11. A functional system has several pieces: Java, OpenNMS, PostgreSQL,
   Tomcat, Xerces, RRD, Ant, etc.  Versionitis is a problem.  On my
   system, some components had to be upgraded because they were all
   of a month old!  The team will bundle things for releases, but
   this problem will make life interesting for developers.

12. Although OpenNMS is designed to be distributed, for now the pieces
   all run on one box.  I think this is a good thing, as it's
   complex enough without being distributed.

Thanks to the team for all the help.  I'll try to actually *use* the
software soon, and I'm happy knowing that friendly expertise is so
accessible.

-- Pete
_______________________________________________
discuss mailing list (discuss@opennms.org)
To subscribe, unsubscribe, or change your list options, go to:
http://www.opennms.org/mailman/listinfo/discuss