How to configure a multi-node UMP cluster to use Unicast instead of Multicast

Products

• DX Unified Infrastructure Management (Nimsoft / UIM)

Issue/Introduction

Summary: 

The UMP portal product normally uses Multicast network communications to locate and communicate additional UMP nodes in a clustered or multi-node environment. This article describes how to configure the product for Unicast instead.

Background:  

In certain circumstances, it may be desirable to set up a clustered or multi-node UMP installation without using multicast.

These circumstances may include:

·        situations where multicast is disabled on a network interface

·        MSP situations where multiple independent multi-node UMP environments may be configured within a single network and should be isolated from each other

·        Other situations where multicast is unavailable (security limitations, OS configurations, etc.)

Environment:  

UMP 8.0 or later

Instructions: 

You will need to follow these steps on both UMP nodes simultaneously - shut down both wasp probes, make the below changes on each node, and then start them up again one at a time (order shouldn't matter.)  If there are more than two nodes, you must perform these steps on all nodes.

1. On each UMP node, locate the following folder: <UIM Install Location>\probes\service\wasp\webapps\ROOT\WEB-INF\classes\
2. Download the attached unicast.xml and copy it to that location on all UMP nodes.
3. Edit the wasp.cfg on both UMP nodes and adjust the "options" key under <startup>, adding any necessary options as follows -- note that there are three spots where you will have to insert an IP address based on your own environment.
   
   options = -Dfile.encoding=UTF-8 -Djava.net.preferIPv4Stack=true -Djgroups.bind_addr=ip.address.of.this.node -Djgroups.tcpping.initial_hosts=ip.address.of.this.node[7800],ip.address.of.other.node[7800] -Djava.server.rmi.hostname=ip.address.of.this.node
   
   If you have more than two UMP nodes, you will need to add their IP addresses to the last option in a similar format, comma-separated, e.g.: 
   
   -Djgroups.tcpping.initial_hosts=ip.address.of.this.node[7800],ip.address.of.other.node[7800],ip.address.of.third.node[7800]
   
4.  Locate the following file on each UMP node and edit it with a text editor:

<UIM Install Location>\probes\service\wasp\webapps\ROOT\WEB-INF\classes\portal-ext.properties

To this file, add the following options (again note that you will need to add an IP address specific to your environment):

cluster.link.autodetect.address=ip.address.of.the.uim.database.server:port

cluster.link.channel.properties.control=unicast.xml 

cluster.link.channel.properties.transport.0=unicast.xml 

ehcache.bootstrap.cache.loader.factory=com.liferay.portal.cache.ehcache.JGroupsBootstrapCacheLoaderFactory 

ehcache.cache.event.listener.factory=net.sf.ehcache.distribution.jgroups.JGroupsCacheReplicatorFactory 

ehcache.cache.manager.peer.provider.factory=net.sf.ehcache.distribution.jgroups.JGroupsCacheManagerPeerProviderFactory 

net.sf.ehcache.configurationResourceName.peerProviderProperties=file=/unicast.xml 

ehcache.multi.vm.config.location.peerProviderProperties=file=/unicast.xml 

Since both UMP nodes can communicate with your SQL database server, specifying its address in the directive above will allow each node to gather some information about which network interface it should be using.   You don't actually have to put the database server's IP here, technically any machine (other than the primary hub) which both nodes can communicate with should work, but using the IP of the database server is usually a guarantee that this will be the case.  Make sure to specify the port as well as the IP.

 In the same file, uncomment the following properties, and then start wasp:
#cluster.link.enabled=true
#lucene.replicate.write=true

 

Now start the wasp probes on all nodes and they should now be running in a clustered fashion but using Unicast instead of Multicast.

Additional Information:

The following line can be added to the portal-ext.properties file to enable additional debug logging around the clustering process:

cluster.executor.debug.enabled=true

The additional debug information will be logged in the portal.log file, found in the wasp folder.

 

Environment

Release: CNMSPP99000-8.2-Unified Infrastructure Mgmt-Server Pack-- On Prem
Component: