For the list of all supported Decision Insight downloads and releases, see the Downloads page.

Topology for HA cluster

This page describes the recommended topology for installing a high availability (HA) cluster. The topology that is described here runs behind a proxy for higher security and uses a 3rd party load balancer (software or hardware) for dispatching client requests.

All nodes must use the same:

  • context root
  • proxy
  • session cookie name

Description

In this environment:

  • The nodes are accessed through a unique proxy.
  • Chosen client browser requests sent to the RNs are dispatched by a load balancer (software or hardware).
    • If a replica node becomes unavailable, it is the responsibility of the load balancer to transfer the requests to another replica.
    • If the HA main node becomes unavailable, it is the responsibility of the load balancer to transfer the requests to the HA backup node.



The IP addresses displayed in the diagram are only examples; you can use any network as long as the IP address for each node.

Limitations

  • The nodes and the proxy/load balancer must use either HTTP or HTTPS, they cannot be mixed.
  • All nodes must use the same host and port in the proxy URL.
  • All nodes must use the same context root.
  • Any new replica node must be declared in the load balancer.

Installation

Install each node (Install a node) by defining the following settings:

  • through the installer or,
  • manually inside the conf/platform.properties file.
HA node Replica node
platform.properties 
com.systar.electron.type=HA
com.systar.electron.host=<Primary host/IP address on internode network, eg: 172.16.5.10>
com.systar.gluon.clusterId=<Generated cluster id, like 00000007-001-000>

In this topology, you must configure proxy URL and context root on the default network:

platform.properties 
com.systar.boson.http.proxyUrl=<Proxy URL on default network, eg: 10.0.0.1>
com.systar.boson.http.contextRoot=<Common context root>

For HTML5 UI, you must configure the session cookie name:

platform.properties 
com.systar.photon.application.sessionCookieName=<Common session cookie name>

To disable the distributed computing feature on the HA nodes:

platform.properties 
com.systar.krypton.distributedcomputing.primaryComputingEnable=false

Database encryption should be shared:

platform.properties 
com.systar.titanium.encryptionKeyFile=${com.systar.platform.conf.dir}/encryption.key
platform.properties 
com.systar.electron.type=REPLICA
com.systar.electron.host=<HA main host/IP address on internode network, eg: 172.16.5.10>
com.systar.gluon.clusterId=<Same cluster id as HA nodes, like 00000007-001-000>

In this topology, you must configure proxy URL and context root on the default network:

platform.properties 
com.systar.boson.http.proxyUrl=<Proxy URL on default network, eg: http://10.0.0.1>
com.systar.boson.http.contextRoot=<Common context root>

For HTML5 UI, you must configure the session cookie name:

platform.properties 
com.systar.photon.application.sessionCookieName=<Common session cookie name>

To disable computings on a replica node:

platform.properties 
com.systar.krypton.distributedcomputing.replicaComputingEnable=false

To totally disable distributed computing, this parameter must be set to false on all replicas.

Database encryption should be shared:

platform.properties 
com.systar.titanium.encryptionKeyFile=<Copy of primary encryption.key>

Example

For an example of the HAProxy configuration for this topology, see Proxy and loadbalancer example with HAProxy.


Related Links