6 Unified Inventory Management Post-Installation Tasks

Enabling Ehcache for the Cluster

To enable ehcache for the cluster, do the following:

1. Ensure you are logged in to the shared disk storage through a command window.

2. Make a backup copy of the ehcache.xml file and rename it to ehcache.xml.cluster.

3. Open the ehcache.xml.cluster file in a text editor.

4. Scroll down to the cacheManagerPeerProviderFactory entry.

5. Change the value of the port entry to any unused port number.

6. Save and close the file.

Enabling Ehcache Manual Discovery

Use the following checklist to verify that manual discovery is the correct setting to use, for peer discovery:

• By default, the peer discovery setting for ehcache is automatic (which employs multi-casting). Multi-cast is known to be unreliable, has bandwidth overload issues and is unsecure. If you feel multi-casting is not the right setting for your environment, perform the steps to enable the ehcache manual discovery setting for the cluster.

• The cacheManagerPeerListenerFactory port defined in the ehcache.xml file shouldn't conflict with any other process running on the same port. Use the netstat command or lsof command to find out which ports are available.

• When the ehcache manual discovery setting is enabled, instead of automatic discovery, the port numbers in the rmi URLS setting should be same as the cacheManagerPeerListenerFactory port defined in the ehcache.xml file.

• In case of single node cluster (i.e different managed servers on the same node), before starting each managed server the cacheManagerPeerListenerFactory port should be modified so that each managed server gets started with a unique cacheManagerPeerListenerFactory port to listen to and avoiding any port binding exceptions.

To enable the ehcache manual discovery setting, do the following:

1. Ensure you are logged in to the shared disk storage through a command window.

2. Open the ehcache.xml file in a text editor.

3. Scroll down to the cacheManagerPeerProviderFactory entry.

4. Comment out the cacheManagerPeerProviderFactory entry.

   Example of automatic discovery entry:

   <cacheManagerPeerProviderFactory
   class="net.sf.ehcache.distribution.RMICacheManagerPeerProviderFactory"
   properties="peerDiscovery=automatic, multicastGroupAddress=230.0.0.1,
   multicastGroupPort=4446"/>
   

5. Add the following entry to the file:

   Example of manual discovery entry:

   <cacheManagerPeerProviderFactory 
   class="net.sf.ehcache.distribution.RMICacheManagerPeerProviderFactory"
   properties="peerDiscovery=manual,
    rmiUrls=//server1:40000/sampleCache1|//server2:40000/sampleCache1|//server1:40000/sampleCache2|//server2:40000/sampleCache2"/>
   

   Note:

   Use the netstat command or lsof command to find an available port.

6. Change the value of the port entry to any unused port number.

7. Save and close the file.

Enabling Ehcache for JGroups

JGroups is a group communication toolkit and has a flexible protocol stack. JGroups extends unicast messaging to provide reliability. If you use JGroups as a replication mechanism, you must download the JGroups related JAR files and make the library additions to the UIM classpath.

To enable the Ehcache for JGroups (UDP + Unicasting), do the following:

1. Ensure you are logged in to the shared disk storage through a command window.

2. Open the ehcache.xml file in a text editor.

3. Scroll down to the cacheManagerPeerProviderFactory entry.

4. Comment out the cacheManagerPeerProviderFactory entry which uses the ”RMICacheManagerPeerProviderFactory”, which by default makes the transport mechanism as multicast.

   Example of automatic discovery entry:

   <!-- <cacheManagerPeerProviderFactory
   class="net.sf.ehcache.distribution.RMICacheManagerPeerProviderFactory"
   properties="peerDiscovery=automatic, multicastGroupAddress=230.0.0.1,
   multicastGroupPort=4446"/> -->
   
   

5. Add the following entry to the file:

   Example of manual discovery entry:

   <cacheManagerPeerProviderFactory
   class="net.sf.ehcache.distribution.jgroups.
   JGroupsCacheManagerPeerProviderFactory"
   properties="connect=UDP(ip_mcast=false;mcast_addr=224.0.0.35;
   mcast_port=45566;ip_ttl=32;mcast_send_buf_size=150000;
   mcast_recv_buf_size=80000):PING(gossip_host=adminserverhost;gossip_port=5555;
   gossip_refresh=15000;timeout=2000;num_initial_members=3):MERGE2:FD_SOCK:
   VERIFY_SUSPECT:pbcast.NAKACK:UNICAST:pbcast.STABLE:FRAG:pbcast.GMS"
   propertySeparator="::" />
   

   Note:

   The protocol stack with UDP and PING as the bottom protocols uses IP multicasting by default to send messages to all members (UDP) and for discovery of the initial members (PING). However, if multicasting cannot be used, the UDP and PING protocols can be configured to send multiple unicast messages instead of one multicast message (UDP) and to access a well-known server (GossipRouter) for initial membership information (PING).To configure UDP to use multiple unicast messages to send a group message instead of using IP multicasting, the ip_mcast property has to be set to false.To configure PING to access a GossipRouter instead of using IP multicast the following properties have to be set:

   • gossip_host: The name of the host on which GossipRouter is started.

   • gossip_port: The port on which GossipRouter is listening.

   • gossip_refresh: The number of milliseconds to wait before refreshing the address entry of the GossipRouter.

   Before any members are started, the GossipRouter has to be started and the administration server is the ideal candidate to be the Gossip Router.

6. Save and close the file.

7. Start the Gossip router, by using the following command:

   java org.jgroups.stack.GossipRouter -port gossip_port -bindaddress gossip_host

   Note:

   The GossipRouter is essentially a lookup service for groups and members. It is a process that runs on a well-known host and port and accepts GET (group) and REGISTER(group, member) requests. The REGISTER request registers a member's address and group with the GossipRouter. The GET request retrieves all member addresses given a group name. Each member has to periodically (gossip_refresh) re-register their address with the GossipRouter, otherwise the entry for that member will be removed (accommodating for crashed members).The property ip_mcast is set to false in UDP and the gossip properties in PING define the GossipRouter to be on the local host at port 5555 with a refresh rate of 15 seconds. If PING is parameterized with the GossipRouter's address and port, then gossiping is enabled, otherwise it is disabled. If only one parameter is given, gossiping will be disabled. Make sure to run the GossipRouter before starting any members, otherwise the members will not find each other and each member will form its own group.

8. Configure the caches in the ehcache.xml file to use JGroupCacheReplicatorFactory. For example:

   <cache
               maxElementsInMemory="10000"
               eternal="true"
               timeToIdleSeconds="0"
               timeToLiveSeconds="0"
               overflowToDisk="false"
               memoryStoreEvictionPolicy="LRU">
               <cacheEventListenerFactory class="net.sf.ehcache.distribution.RMICacheReplicatorFactory"
   properties="replicateAsynchronously=false, replicatePuts=true, replicateUpdates=true, replicateUpdatesViaCopy=true, replicateRemovals=true "/>
   </cache>