This is an alternate version of the task in Section 1, Task 3: Create a Test PDB and Process, Reduce and Report on Your Data. In this version of the task, you update the SPECTRUM database using the SPECTRUM Command Line Interface (CLI). Using CLI, you can enable logging of attributes for more than one model at a time.

1. SAS is installed.

   See Section 1, Task 2: Start Your Data Collection Software.

2. The IT Service Vision server software is installed.

   See Section 1, Task 2: Start Your Data Collection Software.

3. You know the model type name of the devices from which you want to collect data.

   Because SPECTRUM and IT Service Vision are organized based on the model type of a device, in this document logging will be enabled for all devices with the same model type. See Section 1, Task 3: Create a Test PDB and Process, Reduce and Report on Your Data for more information on determining the model type for a particular model.

4. The IT Service Vision server is available on the SPECTRUM management machine.

   See Section 1, Task 2: Start Your Data Collection Software for more information.

All actions for this task are to be performed on the SPECTRUM management machine.

Note: The IT Service Vision file cscsilog contains a shell script that interfaces with the SPECTRUM data base by using the SPECTRUM Command Line Interface (CLI). For more information on CLI, please refer to the SPECTRUM Command Line Interface manual.

1. Customize the cscsilog shell script for updating the SPECTRUM data base to enable data collection.
   
   
   1. Make a copy of cscsilog.

      For the purpose of illustration, assume that !SASROOT=/usr/local/sas/, SPECTRUM is installed in /usr/spectrum, and assume that your work space is /tmp/. In that case, you would use this command:

      cp /usr/local/sas/misc/cpe/cscsilog /tmp/cscsilog

      to copy cscsilog to the /tmp directory.

   2. Customize the cscsilog shell script.

      Near the top of cscsilog you will find a section of user-definable variables that control the execution of the script. These must be customized to work properly at your site.

      • DATA_PATH

        Set this to the SPECTRUM base directory to which SPECTRUM_home refers. For example, if SPECTRUM is installed in /usr/spectrum, DATA_PATH='/usr/spectrum'.

      • DATA_DIR

        Set this to the directory (under the base directory) where the command line interface programs are located. Normally, DATA_DIR='vnmsh'.

      • POLL_INT

        Set this to the number of seconds between polls of the devices. A smaller number will create more network traffic.

      • LOG_RATIO

        Set this to the fraction of polls that are logged. (For instance, if LOG_RATIO is 5, every fifth poll is logged.) This, together with POLL_INT, controls the intervals between data samples.

2. Execute the script

   The script requires at least two parameters. The first is the SPECTRUM landscape (server) name. The remaining parameters are the model types of interest. Many different model types can be enabled for polling and logging at one time. For example, if the SPECTRUM landscape name is sunfin and the model types for data collection are HubCSIMRXi and Rtr_Cisco, then the following should be used to start data collection:

   /usr/spectrum/vnmsh/cscsilog sunfin HubCSIMRXi Rtr_Cisco

3. Watch the output of the script.

   You should see messages similar to the following:

   Starting CLI Local Server ./VnmShd ...
   connect: successful sunfin
   current landscape is 0x400000
   Updating attributes for all models of type Rtr_Cisco
   Id         Name                Value
   0x10071    Polling_Interval    60
   Id         Name                Value
   0x10072    Poll_Log_Ratio      10
   
   Updating attributes for all models of type HubCSIMRXi
   Id         Name                Value
   0x10071    Polling_Interval    60
   Id         Name                Value
   0x10072    Poll_Log_Ratio      10

   where the Id fields are the attribute IDs that correspond to the Polling Interval and Poll_Log_Ratio. These attributes control the rate of polling and archiving of data for the model type.