Integrating IBM Operations Analytics Predictive Insights with Netcool Operations Insights

• Integrate Predictive Insights with a hybrid Netcool Operations Insight deployment.
• Integrate Predictive Insights with Netcool Operations Insight on a full cloud deployment.

1. If you encounter a ports validation error, close Installation Manager.
2. Create the following files:

   touch /tmp/ignorevalidation
   touch /tmp/ignorevalidation.txt
   touch /tmp/ignoreValidation.txt
   touch /tmp/ignoreValidation

     
3. Open Installation Manager and run through the Predictive Insights validation again.
4. When the validation fails on the port-check, a check-box that will allow you to ignore the error and continue is displayed. Select the checkbox, and see that the install completes successfully. Then, apply the latest fix for Predictive Insights - with a tip to create a file named ui_env.properties, based on the ui_env.properties.SAMPLE file in the 'ui' subfolder of the unpacked iFix, checking that the values in the ui_env.properties file are correct for your system.
5. On the server running the Predictive Insights Analytics component, append the following lines to the end of the $PI_HOME/probe/omnibus/probes/linux2x86/stdin-tasp.rules file.

   $Identifier = regreplace($Identifier, "(_\d{13}$)", "" )
   @Identifier = $Identifier
   update(@Identifier)
   
   if (regmatch($TASPIdentifier, "Node")
    || regmatch($TASPIdentifier, "RelatedEvents")
    || regmatch($TASPIdentifier, "Incident")) 
   { 
     discard 
   }

   This ensures that Predictive Insights does not forward Node-consolidated events or RelatedEvents-consolidated events to Netcool Operations Insight. Predictive Insights does not need to forward such events because Netcool Operations Insight already has those capabilities.
6. Ensure that there is a scope-based policy created in Netcool Operations Insight to group events by Node. In the following example the policy is limited to apply to only Predictive Insights events. And the Time Window is set to 20 seconds.

   Note that if you see an "Error fetching field conversions" when you attempt to add the scope-based policy, then install scope-based grouping, as described in this link.

   In a Hybrid deployment, the ServiceDiagnosis tool works well, when opening a Predictive Insights chart from the Event Viewer, but it does not currently work from the new Events page.

   It is advised not to use the Predictive Insights filter in Netcool Operations Insight. It can obscure the Netcool Operations Insight grouping.

Result:

Predictive Insights is successfully integrated with your hybrid Netcool Operations Insight deployment.

For this scenario, events will be forwarded from the stdin probe in on-premises Predictive Insights to Red Hat OpenShift.
Note that this scenario does not include an integration from Netcool Operations Insights in Red Hat OpenShift to the Predictive Insights UI. The Predictive Insights UI will need to be accessed from an on-premises DASH instance.

Complete the following steps on your cloud deployment:

1. Run the following command:

   export RELEASE=noi
   export NAMESPACE=netcool

   Where you can adjust the values for these two environment variables as required for your system.
2. Gather information.

   • Find the ports being used by the Primary and Backup ObjectServers:

     oc get service ${RELEASE}-objserv-agg-primary-nodeport -n ${NAMESPACE}
     NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
     noi-objserv-agg-primary-nodeport NodePort 172.30.252.170 <none> 4100:32172/TCP,30500:30500/TCP 27m
     oc get service ${RELEASE}-objserv-agg-backup-nodeport -n ${NAMESPACE}
     NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
     noi-objserv-agg-backup-nodeport NodePort 172.30.230.163 <none> 4100:31383/TCP,32516:32516/TCP 28m



      
   • Find out what worker nodes are in use by those pods: 

     oc get pod -n ${NAMESPACE} -o wide | egrep "primary|backup"
     noi-ncobackup-0 2/2 Running 0 36m 10.254.20.7 worker2.mySystem.acme.com <none> <none>
     noi-ncoprimary-0 1/1 Running 0 36m 10.254.12.18 worker1.mySystem.acme.com <none> <none>


     Take note that the noi-ncoprimary-0 pod is on the worker1 node and noi-ncobackup-0 is on the worker2

 node.
   • Find the IP address of the worker nodes:

     oc get nodes -o wide | grep worker[12]
     worker1.mySystem.acme.com Ready worker 129m v1.19.0+7070803 10.17.115.111 [...]
     worker2.mySystem.acme.com Ready worker 129m v1.19.0+7070803 10.17.116.195 [...]

      
3. Edit the haproxy configuration file to allow those ports to be accessed from outside the OpenShift system:

   ssh 'root@mySystem-inf.acme.com'

   vi /etc/haproxy/haproxy.cfg

    
   Near the end of the file, add entries like the following example:

       frontend aggp-proxy-primary-port
       bind *:32172
       default_backend aggp-proxy-primary-port
       mode tcp
       option tcplog
   
       backend aggp-proxy-primary-port
       balance source
       mode tcp
       server worker1 10.17.115.111:32172 check
   
       
frontend aggp-proxy-backup-port
       bind *:31383
       default_backend aggp-proxy-backup-port
       mode tcp
       option tcplog
   
       backend aggp-proxy-backup-port
       balance source
       mode tcp
       server worker2 10.17.116.195:31383 check
   
4. Run the following command:

   systemctl restart haproxy
5. Add Predictive Insights columns to the OpenShift primary ObjectServer by running the following command:

   export RELEASE=noi

   oc exec -ti ${RELEASE}-ncoprimary-0 -- bash
   /opt/IBM/tivoli/netcool/omnibus/bin/nco_sql -u root -p $OMNIBUS_ROOT_PWD -s AGG_P
6. Paste the following set of commands into the terminal:

   alter table alerts.status add column TASPAlgorithmInstanceName varchar(50);
   alter table alerts.status add column TASPCorrelationId int;
   alter table alerts.status add column TASPInstanceIdentifier varchar(50);
   alter table alerts.status add column TASPQuantifier varchar(30);
   alter table alerts.status add column TASPUpdateTime time;
   alter table alerts.status add column TASPAnomalyTimestamp time;
   alter table alerts.status add column TASPMetricNameList varchar(1000);
   alter table alerts.status add column TASPMetricScoreList varchar(1000);
   alter table alerts.status add column TASPMetricValueList varchar(1000);
   alter table alerts.status add column TASPMetricResourceNameList varchar(1000);
   alter table alerts.status add column TASPMetricResourceTypeList varchar(1000);
   alter table alerts.status add column TASPMetricNodeList varchar(1000);
   alter table alerts.status add column TASPAnomalousMetrics varchar(1000);
   alter table alerts.status add column TASPMetricGroupNameList varchar(1000);
   alter table alerts.status add column TASPAnomalousResources varchar(1000);
   alter table alerts.status add column TASPAnomalousMetricGroups varchar(1000);
   alter table alerts.status add column TASPParentIdentifier varchar(100);
   alter table alerts.status add column TASPTopic varchar(50);
   alter table alerts.status add column TASPDirection varchar(50);
   alter table alerts.status add column TASPRelatedID varchar(100);
   alter table alerts.status add column TASPActualValue varchar(30);
   alter table alerts.status add column TASPExpectedValue varchar(30);
   alter table alerts.status add column TASPAttributeNames varchar(1000);
   alter table alerts.status add column TASPAttributeValues varchar(1000);
   alter table alerts.status add column TASPIdentifier varchar(100);
   alter table alerts.status add column TASPPriorityScore varchar(100);
   alter table alerts.status add column TASPAlarmNote varchar(1000);
   
   go

   

   Confirm the columns were added by running the following command:

   describe alerts.status

   go
   Then, run the quit and exit commands.
7. Add Predictive Insights columns to the OpenShift backup ObjectServer by running the following command:

   oc exec -ti ${RELEASE}-ncobackup-0 -- bash
   /opt/IBM/tivoli/netcool/omnibus/bin/nco_sql -u root -p $OMNIBUS_ROOT_PWD -s AGG_B

   Then, repeat step 6 for the backup
 ObjectServer.
8. On the server running the Predictive Insights Analytics component, append the following lines to the end of the $PI_HOME/probe/omnibus/probes/linux2x86/stdin-tasp.rules file.

   $Identifier = regreplace($Identifier, "(_\d{13}$)", "" )
   @Identifier = $Identifier
   update(@Identifier)
   
   if (regmatch($TASPIdentifier, "Node")
    || regmatch($TASPIdentifier, "RelatedEvents")
    || regmatch($TASPIdentifier, "Incident")) 
   { 
     discard 
   }

   This ensures that Predictive Insights does not forward Node-consolidated events or RelatedEvents-consolidated events to Netcool Operations Insight. Predictive Insights does not need to forward such events because Netcool Operations Insight already has those capabilities.
9. Ensure that there is a scope-based policy created in Netcool Operations Insight to group events by Node. In the following example the policy is limited to apply to only Predictive Insights events. And the Time Window is set to 20 seconds.

   Note that if you see an "Error fetching field conversions" when you attempt to add the scope-based policy, then install scope-based grouping, as described in this link.

   It is advised not to use the Predictive Insights filter in Netcool Operations Insight. It can obscure the Netcool Operations Insight grouping.