Trying to get the hang of OpenNMS

*Hello all,

I have been evaluating OpenNMS and I have to say that the learning curve is
somewhat steep (terminology, configuration tools and style). I took some
notes of my first experiences, and will be including them below. Any hints
and pointers will be appreciated where I have misunderstood things. Did I
understand these things correctly?

I understood the categories for Dashboard will be configured by editing
files. Fine, but let's take the default Dashboard... I have a few
automatically discovered servers. How can one assign them to "Servers" and
"PROD"? I was expecting a drop down menu in asset information or something
similar. Something added my discovered servers to the main window
categories as well, but I have no idea what. After couple hours of research
I haven't found any way to accomplish moving the servers around.

What views (Dashboard etc) have automatic updating, and what must be
reloaded by users to update the visible information? This should be more
obvious from the user interface.

How to configure SSH monitoring using keys (in case username/password
logins are disabled)? I would expect this to be available by default.
However I can see only username/password fields in the configuration form.

Performance data can't be collected by default by using SSH remote
commands? That might be good for the environments without SNMP but Linux
everywhere. Running vmstat/iostat/whatever remotely would provide decent
monitoring capabilities. Is this available somehow already, with the
current SSH poller?

There is a surprising amount of editing configuration files. For instance
when getting SNMP monitoring running. As the work is mostly simple mibObj
node additions to XML configuration files, imho a browser interface should
be provided for managing them. It wouldn't be specifically hard to
implement but it would make implementing SNMP with OpenNMS less
intimidating. Or ship with default MIBs for most common stuff - basic
RedHat Linux, Windows Server 2012 and such...

Next thing I started looking at was monitoring HTTP services. Seeing that
the port is listening is barely enough, some basic sanity checking should
obviously be done. The PSM is fine for extranets and public services but it
seems to be useless for any corporation internal services because it
doesn't require SPNEGO. After all it is the only widely implemented method
for SSO (typing the username/password repeatedly to services isn't really
SSO) and I know of environments where no end user actually even knows their
username and password (PKCS#11 + Kerberos all the way - they have the card
and the PIN for it - and they need to use the PIN also only when they login
for the first time).

All the PSM is able to see when SPNEGO is required is nasty HTTP 401. I
haven't seen one single service in internal environments for 5-10 years
that wouldn't have required SPNEGO so I was kind of surprised about this
deficiency. So I went for GpPlugin and some scripting, took two hours or so
to hack something similar to PSM but with Spnego support. It works near
perfectly but the fact that I had to develop myself some so rudimentary
support makes me feel a bit bad about OpenNMS...

Selenium would work better as it automates Firefox (which definitely does
transparently SPNEGO), but in my opinion Selenium kind of comes with a
heavy package if you need it only for the SPNEGO. That's kind of why I had
expected PSM to support it.

Afaik Java has APIs for loading keytabs, initializing the gssapi context
using it, and responding the to the Negotiate challenge. It shouldn't be a
major work to add that support to PSM, but the requirement of proper test
environment and testing is laborious.

JDBC monitoring is solid the way it is implemented. Configuration is
slightly intimidating once more but I like the implementation. It allows
actual sanity checking which is very very good. (Especially Oracle
databases sometimes allow connections but for any actual SQL queries they
give error messages. What OpenNMS offers is very good for monitoring such
products.)

I like the syslog receiving feature. Most likely could pipe some server
errors (dmesg stuff) there.

Is there a tutorial kind of documentation about how to integrate OpenNMS
with a ticketing system that is not currently supported? Probably involves
implementing something towards org.opennms.api.integration.ticketing.* and
again XML file configuration? If someone knows of a good howto I would be
interested in taking a look at it.

In overall I like what I have seen. I haven't bumped into some single thing
that could be categorized straight as bug, and the products works. The
discovery approach is much better than configuring every single monitored
target manually. (I read about the design principles from the Wiki and I
understand the approach. It's a good one.) Scalability seems to be there,
as well as remote monitoring support. I like the user interface as well,
but would wish more "create a jsp and plug it in" kind of extendability.
*
Uhh that's how far I got in ~6 hours. :)

**
*
*
**

------------------------------------------------------------------------------
The Go Parallel Website, sponsored by Intel - in partnership with Geeknet, 
is your hub for all things parallel software development, from weekly thought 
leadership blogs to news, videos, case studies, tutorials, tech docs, 
whitepapers, evaluation guides, and opinion stories. Check out the most 
recent posts - join the conversation now. http://goparallel.sourceforge.net/

_______________________________________________
Please read the OpenNMS Mailing List FAQ:
http://www.opennms.org/index.php/Mailing_List_FAQ

opennms-install mailing list

To *unsubscribe* or change your subscription options, see the bottom of this page:
https://lists.sourceforge.net/lists/listinfo/opennms-install