9.7. Installing and Configuring MCollective on Node Hosts

The broker host uses MCollective to communicate with node hosts. MCollective on the node host must be configured so that the node host (Host 2) can communicate with the broker service on Host 1.
In a production environment, two or more messaging hosts would typically be configured on machines separate from the broker to provide high availability. This means that if one messaging host fails, the broker and node hosts can still communicate.

Procedure 9.5. To Install and Configure MCollective on the Node Host:

  1. Install all required packages for MCollective on Host 2 with the following command: 
    # yum install openshift-origin-msg-node-mcollective
  2. Replace the contents of the /opt/rh/ruby193/root/etc/mcollective/server.cfg file with the following configuration. Remember to change the setting for plugin.activemq.pool.1.host from broker.example.com to the host name of Host 1. Use the same password for the MCollective user specified in the /etc/activemq/activemq.xml file on Host 1. Use the same password for the plugin.psk parameter, and the same numbers for the heartbeat parameters specified in the /opt/rh/ruby193/root/etc/mcollective/client.cfg file on Host 1: 
    main_collective = mcollective
collectives = mcollective
libdir = /opt/rh/ruby193/root/usr/libexec/mcollective
logfile = /var/log/openshift/node/ruby193-mcollective.log
loglevel = debug

daemonize = 1
direct_addressing = 0

# Plugins
securityprovider = psk
plugin.psk = asimplething

connector = activemq
plugin.activemq.pool.size = 1
plugin.activemq.pool.1.host = broker.example.com
plugin.activemq.pool.1.port = 61613
plugin.activemq.pool.1.user = mcollective
plugin.activemq.pool.1.password = marionette

plugin.activemq.heartbeat_interval = 30
plugin.activemq.max_hbread_fails = 2
plugin.activemq.max_hbrlck_fails = 2

# Node should retry connecting to ActiveMQ forever
plugin.activemq.max_reconnect_attempts = 0
plugin.activemq.initial_reconnect_delay = 0.1
plugin.activemq.max_reconnect_delay = 4.0

# Facts
factsource = yaml
plugin.yaml = /opt/rh/ruby193/root/etc/mcollective/facts.yaml
  3. Configure the ruby193-mcollective service to start on boot: 
    # chkconfig ruby193-mcollective on
  4. Start the ruby193-mcollective service immediately: 
    # service ruby193-mcollective start

    Note

    If you use the kickstart or bash script, the configure_mcollective_for_activemq_on_node function performs these steps.
  5. Run the following command on the broker host (Host 1) to verify that Host 1 recognizes Host 2: 
    # oo-mco ping

9.7.1. Facter

The Facter used by MCollective is an important architectural component of OpenShift Enterprise. Facter is a script that compiles the /opt/rh/ruby193/root/etc/mcollective/facts.yaml file, and lists the facts of interest about a node host for inspection using MCollective. Visit www.puppetlabs.com for more information about how Facter is used with MCollective. There is no central registry for node hosts, so any node host listening with MCollective advertises its capabilities as compiled by the Facter.
The broker host uses the facts.yaml file to determine the capabilities of all node hosts. The broker host issues a filtered search that includes or excludes node hosts based on entries in the facts.yaml file to find a host for a particular gear.
The Facter script runs on the node host in one minute increments, which can be modified in the /etc/cron.minutely/openshift-facts cron job file. You can also run this script manually to immediately inspect the new facts.yaml file.