Added monitoring with nagios

[gildub]

This new version is using an external puppet nagios module.
Which makes more sense since this is not directly related with openstack anyway.
This has the advantage to reduce impact on quickstack as much as possible.

This add a 'Nagios hostgroup' and monitoring::server and monitoring::client to wrap calls to the external nagios module.

This is impacting only non HA to start with (Neutron/Nova_network controllers/compute/networker)

Tested with latest RHOS5/RHEL7

This requires:

• BZ#1118537 - Nagios plugins (load and disk) to be added to RHOS repos
• RFE - Nagios module for Openstack openstack-puppet-modules#83

[jguiditta]

Visually this looks fine, though I cannot test until the mentioned dependent BZ is completed so I have the right packages

[gildub]

@jguiditta, the nagios plugins disk and load are available in the RHOS-5-RHEL-7 repo.
Looks like the BZ#1118537 is not up to date.

[jguiditta]

Since this is solving a BZ, could you please update the commit message to be similar to the below? thanks, will report test results later today separately.

BZ #1040237 - Quickstack puppet classes for OpenStack monitoring (Nagios).

https://bugzilla.redhat.com/show_bug.cgi?id=1040237

This adds a 'Nagios hostgroup' and monitoring::server and monitoring::client to wrap calls to the external nagios module. This impacts only non HA to start with (Neutron/Nova_network controllers/compute/networker)

[jguiditta]

Missing commas, this gives the error:

[jguiditta]

Ok, I needed to update concat from src as well, which got me to 'mostly working'. First error is:

Warning: Firewall001 nagios incoming: Unable to persist firewall rules: Execution of '/usr/libexec/iptables/iptables.init save' returned 1:

This is likely because you ned to pull in quick stack::openstack_common, which sets up the basic firewall config.

The second error is a missing package:

Error: Execution of '/usr/bin/yum -d 0 -e 0 -y list nagios-plugins-http' returned 1: Error: No matching Packages to list

So, as I see it, couple minor fixes needed here:

1. Include openstack_common
2. change default to not have nagios enabled
3. get ppm rpms builds updated so this will work
4. get the missing nagios http rpm in the proper place (I'll have to double check tomorrow to make sure my nodes were not just pointed at an older repo that is missing that rpm)

Other than those minor bits, this is looking good.

[jguiditta]

Verified I am using the latest rhos repo, and that it doesn't not have any nagios+http rpm. So, the above 4 issues, plus the request for updated commit message, and this will likely be ready (though I would like to be able to actually test it, not able to completely do so yet due to the above errors).

[gildub]

@jguiditta, thanks for your feedback

• Added Doc text to BZ #1040237
• Sorry about missing comas (rebasing!): done
• Makes sense to have monitoring deactivated by default (''): done

1. The firewall issue:
   Moving rules to quickstack::monitoring::client and server classes.
2. The missing nagios-plugins-http package issue:
   It seems related to the nagios-plugins-* being renamed monitoring-plugins-*. This is an ongoing process, see RHBZ#1054340.

I'm patching puppet nagios module accordingly for 1. and 2. Will provide update.

[gildub]

Be aware this is very confusing:
Both Fedora20 Everything and EPEL7 uses nagios-plugins
but RDO/Icehouse uses monitoring-plugins: https://repos.fedorapeople.org/repos/openstack/openstack-icehouse/epel-7/

Since the puppet modules is outside of quickstack scope, the nagios-plugins modules which are still available everywhere on RHEL/CentOS/6/7 and Fedora20 will be used.
The puppet modules is not going to be changed.

[gildub]

Pull Request for OPM to integrate latest puppet-nagios-openstack for removed FW rules:
redhat-openstack/openstack-puppet-modules#107