[Xymon] terabithia vs self compiled

[Galen Johnson]

Thanks, JC.  I figured it would be FHS (I do so love the simplicity of /usr/local, though).  I'm going to have to refactor git and ansible (fortunately not fully baked yet) to accommodate this but it should be worth the effort to be able to just 'yum update'.

Thanks for clarifying "local".   Not my setup but I can see how it is helpful.

=G=
________________________________________
From: J.C. Cleaver <cleaver at terabithia.org>
Sent: Tuesday, February 9, 2016 6:17 PM
To: Galen Johnson
Cc: xymon at xymon.com
Subject: Re: [Xymon] terabithia vs self compiled

On Tue, February 9, 2016 2:36 pm, Galen Johnson wrote:
> Hey,
>
>
> I'm considering switching over to using the RPMs that JC provides through
> Terabithia rather than compiling from scratch.  I'm assuming it should as
> simple as just copying over the RPMs and yum installing them.  Any
> gotcha's to look out for?  Just want to be sure that my tweaks to the
> various configs will work and the best way to handle the client tests.  I
> use include and directory a fair bit.

For a general overview of the changes, see
http://terabithia.org/rpms/xymon/xymon.README.terabithia. I'd definitely
advise testing first.


The biggest immediately-visible difference between the two is that
everything's in FHS locations in the RPM, and $XYMON(CLIENT)HOME is
de-emphasized throughout. On the client the RPM version of xymonclient.cfg
does an "include" on /etc/sysconfig/xymon-client to bring in the
destination server (which on a msgscache-enabled box would be 127.0.0.1).
If you're deploying straight over an existing install your xymonclient.cfg
won't be replaced, so you'll need to bring that line in by hand. (This is
somewhat inverted from the runclient.sh logic.)


include and directory should work as expected for the rest of your
configurations. The RPM creates ".d" directories for most of the configs
and includes them. I tend to leave the main configs untouched.


On the main xymon server, you'll want both the "xymon" and "xymon-client"
RPMs. The "local client" on the main server is handled with an include in
xymonserver.cfg to the xymonclient.cfg file, which is then handled like
normal. (This lets your main xymon server report its own data to multiple
locations like any other client.)


> I've never exactly been clear on 'local' mode but the majority of my
> monitored systems use msgcache/pulldata so I'm not sure which client rpm
> makes the  most sense.

"local" mode is when thresholds are being evaluated on the client itself
instead of by your main xymond server. This means the client is taking the
raw data and sending it to a local copy of xymond_client to, using your
configs in localclient.cfg, turn that into red/yellow/green status
messages for the xymon server (for example, for CPU load).


In xymon's "server" mode (which is the default), all of this processing
happens centrally according to rules specified at the xymond server in
analysis.cfg.


If you're running a client in "local" mode using the RPMs, you'll also
want the "xymon-client-local" package on top of "xymon-client". (It's
broken up this way because xymond_client needs PCRE, so that's another
dependency.)


> Also, is there a repository definition I can add that will let me just run
> 'yum update' and it can bring in the updated rpms?  I didn't see anything
> obvious on the terabithia site (or I could just be blind).

This should be what you need:
http://terabithia.org/rpms/xymon/terabithia-xymon.repo


HTH,

-jc