Nokia NSP server redundancy

Two Nokia NSP servers can run in a redundancy pair (that is, one runs as the primary server and the other as a backup server). This affects the way that you configure the probe.

If the primary NSP server is down while the probe is connected, then the probe will attempt to connect to the secondary NSP server which will take over the primary server role. The probe cannot connect to the secondary NSP server if the primary server is still operational.

To enable fail-over support, use the following steps:

Change probe properties as below:
```
RotateEndpoint: "true"
RetryCount = 100
RetryInterval = 10
```
RotateEndpoint required to be used together with RetryCount so that it can be fail-over to the secondary server settings during the retry/reconnect attempts.

Edit the nokiaNspKafkaTransport.properties to point to the nokiaNspRestMultiChannelHttpTransportFailover.json file which contains the fail-over primary and secondary server settings:

#==============================================================================
# LOCATION OF FILE CONTAINING REST CONNECTION PROPERTIES
#==============================================================================
httpConnectionPropertiesFile=/opt/IBM/tivoli/netcool/omnibus/java/conf/nokiaNspRestMultiChannelHttpTransportFailover.json

Edit the nokiaNspRestMultiChannelHttpTransportFailover.json file to configure the primary and secondary server IP/hostname and port.
```
"FailOverServer":
      {
         "Primary":
         {
            "authenticationServer": "10.0.0.1:443",
            "restAPIServer": "10.0.0.1:443"
         },
         "Secondary":
         {
            "authenticationServer": "10.0.0.2:443",
            "restAPIServer": "10.0.0.2:443"
         }
      }
```
Check with Nokia NSP regarding the port if there is any changes that not using the default port in the deployment.

Leave the ++authenticationServer++ and ++restAPIServer++ as it is in the configuration file. The probe will replace these variables with your defined FailOverServer.Primary|Secondary.authenticationServer|restAPIServer during runtime.

Edit the $OMNIHOME/java/conf/nokiaNspKafkaConnectionProperties.json file for Nokia NSP Kafka brokers configuration:

{
    "zookeeper_client" :
        {
            "target" : "<Nokia SDN 1 Host IP>:<Zookeeper port>,<Nokia SDN 2 Host IP>:<Zookeeper port>",
            "properties" : "",
            "java_sys_props" : "",
            "topic_watch": true,
            "broker_watch": true
        },
    "brokers" : "SSL://<Nokia SDN 1 Host IP>:<Kafka port>,SSL://<Nokia SDN 2 Host IP>:<Kafka port>",

    "topics": "",
    "kafka_client" :
        {
            "properties" : "<Absolute Path To $OMNIHOME>/java/conf/nokiaNspKafkaClient.properties",
            "java_sys_props" : ""
        }
}

Note: For SSL, the default Kafka port is 9192, and the zookeeper port is 2281. For plaintext, the default Kafka port is 9092, and the zookeeper port is 2181.

If the Nokia SDN is deployed with high-availability (HA), the host IP here will be referred to the HA's virtual IP.

Edit $OMNIHOME/java/conf/nokiaNspRestMultiChannelHttpTransportFailover.json under GLOBAL properties to configure errorResponseLimit and failOnHTTPResponseCodes. You must configure failOnHTTPResponseCodes to react to a response error of 307, 500, and 503, because these error codes indicate that the probe is not connected to the correct active server or that the NSP server is not yet ready. Thus when probe received these errors, the probe should fall back to retry the connection.

   "GLOBAL":
   {
      ...
      "errorResponseLimit": "5",
      "failOnHTTPResponseCodes": "307,500,503",
      ...