Re: Generation of distributed nagios configuration w/cfengine

[Christian Pearce]

I just read the INSTALL.  There is a lot to digest.  I want to take some
time this afternoon and play with it.  My initial reaction is that this
is very complex and a little overwhelming to a new reader.  But I think
that could be improved by having a better overview.  Let's face it
explaining this stuff is tough. (Or maybe it is to early in the
morning)  That was offered as a suggestion and not a harsh criticism. 
Here are some other thoughts.

1. Ruby? Ugh (Just kidding, I need an excuse to learn it)
2. module:nagios in subversion is not a good filename, I can't click on
it without getting "module is not a registered protocol.
3. Why do you push the a remote.cfg out to the nagios clients.  Is this
for people who don't use Cfengine?
4. I am not certain of the integration of Cfengine (I need to reread
INSTALL)
6. Can we setup a mailing list? I don't think it would be appropriate to
hash out problems about naginator?
7. This opens up an opportunity to start discussing the idea of a
Cfengine Module Repository. What we need is a set of guidelines for
writing modules.  This would help with plugging in any module and
getting it to work in any Cfengine environment.  Provided the Cfengine
implementor implemented the guidelines correctly.  I am not certain if
this is possible.  But think of the merit?  We can start sharing the
wealth of Cfengine knowledge beyond just talking about it.  This would
provide great leverage.  This is the first project I have seen that
would tap into the possibility
8. I want to help.

On Mon, 2004-08-02 at 02:14, Luke A. Kanies wrote:
> Hi all,
> 
> I've been working on this project for a while, and I think I've finally 
> reached 
> what I would call a beta.
> 
> I've now got a collection of scripts (all written in Ruby) which generate the 
> necessary configurations for running an appropriate nagios configuration on 
> each node (using nsca to push updates to the central server) plus an 
> aggregated 
> configuration for the central node.  This system is completely 
> self-maintaining, built off the information I store in my cfengine 
> configurations.  Using cfengine, it automatically restarts the nagios 
> processes 
> when configurations change, and it automatically builds and redistributes the 
> configurations as necessary.
> 
> As with all betas, the documentation is not yet up to snuff, but you can find 
> the (current) main page at:
> 
> http://madstop.com/naginator
> 
> There you'll also find a tarball of the latest snapshot (as of today) plus a 
> link to the Subversion code.  Install instructions are available at 
> http://madstop.com/svn/naginator/INSTALL, along with mucho detail about the 
> whys and wherefores.
> 
> As with all solutions, this one works for the author (me), but others might 
> find it useful, too.  I do need to point out, though:  This is not 
> necessarily 
> terribly easy to set up, as its setup is meant to be entirely automated with 
> a 
> tool like cfengine.  The focus of its development is on its ability to 
> maintain 
> itself over time; as I add services to my hosts (using cfengine), I want to 
> automatically start monitoring them, and as I add things like disks, I want 
> them to automatically start being monitored.
> 
> Yes, I have achieved a completely self-maintaining monitoring system with 
> Nagios.  It still has some quirks, but hopefully if a few more people use it, 
> we can iron them out and get something good.
> 
> I will gladly take suggestions and recommendations.  I'm especially 
> interested 
> in people who might be integrating this with their own larger management 
> system, as it will force me to abstract out the cfengine parts.
> 
> On the other hand, if anyone is interested in how this integrates with the 
> rest 
> of my automation, please email me off-list.
> 
> This package also includes a full, LALR parser for nagios configuration 
> files. 
> It's written in ruby, but it makes parsing files trivial, and it includes 
> full 
> object definitions for all currently known Nagios objects.
> 
> I also have a generated LDAP schema for nagios objects, although I'm not 
> currently using it.  Please let me know if you're interested.
> 
> Thanks,
> Luke
-- 
Christian Pearce
http://www.commnav.com
http://www.perfectorder.com