[Xymon] vagrant example of xymon-4.3.19 server (rhel6) + 3 clients (rhel6, rhel7, ubuntu14)

[J.C. Cleaver]

On Sun, May 3, 2015 4:38 pm, Marcin Dulak wrote:
> Hi,
>
> the example is available at
> https://github.com/marcindulak/vagrant-xymon-tutorial
> I hope Xymon's beginners will find it useful.

Hi Marcin,

Thanks for your report! Definitely the more documentation the better :)

>
> In the process of creating it I noticed some problems with the
> documentation
> https://www.xymon.com/help/install.html:
> 1. RHEL6/7 documentation is missing dependencies: yp-tools c-ares-devel
> 2. Debian documentation is missing dependencies: libc-ares-dev dpkg-dev

For c-ares-devel, the default tarball includes a bundled copy of c-ares,
however since it's generally preferable to use shared libraries when
possible I agree it should be called out for package installation. It's
unfortunate that RH placed it into the 'optional' channel, as that makes
it a bit more difficult to add into a managed environment for building.

I'm curious where the dependency for yp-tools came in, however. On EL6 and
EL7, that should not be a req.


> 3. I believe fping is no longer necessary as a dependency when building
> RPMS or debs

It's not a dependency in the current build, but it is very strongly
recommended. fping3 has been available for a while now and is quite
stable. xymonping should be considered deprecated in favor it.

> 4. The /etc/init.d for xymon-client seems still missing on Debian/Ubuntu
> as reported here:
> http://comments.gmane.org/gmane.comp.monitoring.hobbit/30534
>
> Best regards,
>
> Marcin


Thanks,

Yes, especially for systemd systems, there's a definite gap there. I've
actually gotten it working, but with some significant runtime
modifications to the present tarball (some of this was also necessary for
SELinux compliance).

I'd been working to split these patches apart to allow for an orderly,
feature-at-a-time commit process into SVN but it's been proving more
difficult than expected to separate the modifications while still ensuring
that intermediate stages actually build. It might be easier to simply do a
mass patch to bring everything in and then enter a stabilization period
for beta testing of a 4.4 branch.

If I go with that approach, systemd and a more normalized initscript
approach should be working sooner rather than later.


Regards,

-jc