NerveCenter^TM: Downstream Alarm Suppression - Contents - Testing the Model -
Understanding How the Model Works      Understanding the Technical Details

Testing the Model

The alarm suppression model is based on this concept: by monitoring whether nodes are dependent on other nodes (parent-child relationships) and by keeping each node's status updated proactively, the model can make accurate assessments as to what the statuses of dependent, or child, nodes are.

The following sections describe how to test the models:

• Importing the New Model
• Identifying Parent-Child Relationships
• Making the Relationship Information Available to NerveCenter
• Testing the Alarm Suppression Model
• Running Node Availability Reports

Importing the New Model

The new alarm suppression model is not included in the default database that is installed with NerveCenter v3.8. You must import the model before you can use it. The model includes all of the objects you need, including the alarms, polls, masks, Perl subroutines, and so on.

---

   Older versions of the imported objects will be overwritten (for example, IcmpPoll).

---

 
  To import the new model:

1. Start the Client and connect to the NerveCenter Server.
2. From the Server menu, select Import Objects and Nodes.
3. Select Browse.
4. Double-click the node_status directory.
5. Select nodestatus_dwnstrm.mod and then Open.
6. Select OK.

A message is displayed when the file has been imported. The following alarms are listed in the alarm list:

• DwnStrmIcmpStatus
• DwnStrmIcmpStatus_LogToDB
• DwnStrmSnmpStatus
• DwnStrmSnmpStatus_LogToDB

See Designing and Managing Behavior Models for complete details on importing models.

Identifying Parent-Child Relationships

In order to use NerveCenter's Downstream Alarm Suppression behavior model, it is necessary to establish the parent-child relationship between nodes. You can let OVPA extract relationship information from HP OpenView and either store it in the NerveCenter database or in a text file. You can also create the text file manually.

---

   By default, OVPA does not get information about a node's parents from your network management platform. You must configure OVPA to collect that information by doing the following steps.

---

 
  To identify parent-child relationships using OVPA:

1. Make sure HP OpenView is running. Also make sure the NerveCenter Server is running.
2. Make sure that your network management platform is set up as your node source in the NerveCenter Administrator.
3. If OVPA is running, stop it by typing ovstop ovpa at the command line.
4. Start OVPA in parenting mode from the command line by typing one of the following commands:
   • ovpa -pc

     OVPA runs and computes parenting information, resychronizing the information periodically. The how often OVPA resychronizes information is configurable through the Node Source tab in the NerveCenter Administrator. The default resync parent rate is 600 seconds.

   • ovpa -pc -writeParentsToFile hostname

     hostname is the name of the machine on which the NerveCenter Server runs. OVPA computes the parenting information, writes it to a file named hostname_PC.dat, and then stops.

 
  To change the resync parent rate:

1. Open NerveCenter Administrator and connect to the appropriate NerveCenter Server.

   For further instructions, see Managing NerveCenter.

2. Select the Node Source tab.

   NerveCenter displays the Node Source tab.

   

3. In the Resync Parent Rate field, type the number of seconds you want between each resync attempt.

   If left blank, the default resync parent rate is 600.

4. Select Save.

See Integrating NerveCenter with a Network Management Platform for complete details about starting and stopping OVPA and the NerveCenter Server, as well as instructions for setting up a node source.

 
  To identify parent-child relationships manually:

1. Open a new text file.
2. Include a line for each node that has parents. Use the following syntax:
       child parent

   where child is the name of the node and parent is the name of each node on which the child is dependent. If you have more than one parent, separate parents by typing a space between each one.

   ---

      If NerveCenter uses a full domain name for the node, use the full name in this file to refer to that node.

   ---

   For example, if nodeA is dependent on nodeB.domain.com and nodeC, and nodeB.domain.com is dependent on nodeD, then the contents of the text file would look like this:

   
       nodeA nodeB.domain.com nodeC
    nodeB.domain.com nodeD
3. Save and close the file.

   The name and location of the file do not matter, as long as you remember the location to use in the next procedure.

Making the Relationship Information Available to NerveCenter

If you created a text file with the relationship information -- either manually or by using OVPA -- you must load that information into NerveCenter.

 
  To load relationship information into NerveCenter:

1. In the NerveCenter Client, create an alarm that you can transition on demand. On the transition, call a Perl subroutine that includes the following function:
       NC::LoadParentsFromFile(FileName);

   where FileName is the name of the file you created.

2. Transition the alarm. After the alarm transitions, you can turn the alarm off.

   ---

      If you modify the file, you must repeat this procedure.

   ---

To make sure the contents of the file were read correctly, you can create another alarm with a Perl subroutine that includes the following function:

The information will be written to the file on the local machine.

To remove relationship information, you can create an alarm with a Perl subroutine that includes the following function:

Testing the Alarm Suppression Model

You can test the model by turning the DwnStrmSnmpStatus alarm (see DwnStrmSnmpStatus Alarm) and DwnStrmIcmpStatus alarm (see DwnStrmIcmpStatus Alarm) on, and then simulating a node being unreachable.

 
  To test the alarm suppression model:

1. Make sure the Client is connected to the NerveCenter Server.
2. From the Admin menu, select Alarm Definition List.

   The Alarm Definition List dialog is displayed.

3. In the listbox, right-click on DwnStrmSnmpStatus and select On.
4. In the listbox, right-click on DwnStrmIcmpStatus and select On.

The DwnStrmSnmpStatus and DwnStrmIcmpStatus alarms monitor the status of the managed nodes.

---

   To use the model to log data against which you can run availability reports, use DwnStrmSnmpStatus_LogToDB and DwnStrmIcmpStatus_LogToDB instead of the versions that don't log data (DwnStrmSnmpStatus and DwnStrmIcmpStatus).

---

To simulate a node being unreachable, you can change the IP address of an existing node or that of a new node to an invalid address for your network. (For example, you might use 10.10.10.10.) If you have a test network available, you can also make nodes unreachable by unplugging devices -- for example, a router.

NerveCenter detects errors since the node no longer responds to polls. As a result, NerveCenter reevaluates and updates the status of the node. If the alarm is in an AgentDown, DeviceDown, or Unreachable state, NerveCenter suppresses suppressible alarms for that node until it is available again.

To make sure the statuses of the nodes are correct, you can create an alarm with a Perl subroutine that includes the following function:

The information will be written to the file in the NerveCenter installation directory on the local machine.

Running Node Availability Reports

If you turned on the alarms that log transition data, you can run node availability reports against that data.

---

   Currently, these reports are available for NerveCenter Servers running on Windows only. Also, you must have imported the model and turned the correct alarms on: DwnStrmIcmpStatus_LogToDB and DwnStrmSnmpStatus_LogToDB.

---

If you are running NerveCenter on UNIX, you can export your data to Windows and run the reports there using Crystal Reports, or use another third-party reporting utility on UNIX.

 
  To run the node availability reports:

1. From the Admin menu in the NerveCenter Client, select Report List.

   The Report List dialog box is displayed.

2. Select New.

   The Add Report dialog box is displayed.

   Add Report Dialog Box

   

3. In the Report Select List, select one of the following reports:
   • availsum.rpt -- Lists each node and percent availability by property group. This report offers availability information at a glance.
   • availstat.rpt -- Lists each node and the amount of time it spent in each state, as well as the overall outage and availability. This report includes both actual times and percentages of time.
   • availtrans.rpt -- Lists each node and its state transitions. If you have a large number of nodes, this report can be quite long.
4. In the Report Name field, type a name for your report (optional).
5. In the Report Author field, type your name (optional).
6. In the Description field, type any information that will help you or others understand the report or why it was generated (optional).
7. Select the Override Server Name in Report checkbox.
8. Select OK.

   The report is added to the report list.

9. Select the report, and then select Run.

The report is generated and displayed. availstat.rpt Reportis an example of the availstat.rpt report.

availstat.rpt Report

See Monitoring Your Network for more details about running and viewing reports.