Consider a situation that our operator would like to be aware of before customers begin complaining.  It is with these situations in mind we should consider provisioning thresholding in NMIS.

Related Information

NMIS Threshold Configuration

Basic and Advanced Thresholds in NMIS8

Provisioning Thresholding

Thresholding can be accomplished with the following steps:

  • What data is being collected which can be thresholded?
  • Add a threshold property to the model section.
  • Add threshold values to the Common-threshold.nmis file.
  • Add statistics extraction (pulling data from the RRD file and formating it) to the Common-stats.nmis file.
  • Test the newly provisioned thresholding policy.
  • Consider advanced threshold using controls (Adding boolean logic and regular expressions for precision selection).

Consider the following questions:

  • What would you like to threshold?
  • How feasible is the thresholding candidate?
  • Can the metrics be reduced/translated/combined into a meaningful threshold?
  • What should the corresponding event name for the threshold be?
    • The event name must include "Proactive" at the beginning in order for NMIS to process it correctly.  e.g. "Proactive Temp" or Proactive CPU Load".

Steps and Examples

#1.  Add a threshold property to the model section.

#2.  Add threshold values to Common-threshold.nmis

#3.  Add statistics extraction to Common-stats.nmis

 

Testing Thresholds

After provisioning a threshold it is important to verify it is working as intended.  Run the threshold manually as below.

usr/local/nmis8/bin/nmis.pl type=threshold debug=true

### Look for output similar to the following that displays the current value and the threshold value.

22:27:36 getThresholdLevel, Start theshold=ssCpuRawUser, index=
22:27:36 getThresholdLevel, found threshold=ssCpuRawUser entry=default
22:27:36 getThresholdLevel, threshold=ssCpuRawUser, item=ssCpuRawUser, value=2.15
22:27:36 getThresholdLevel, result threshold=ssCpuRawUser, level=Normal, value=2.15, thrvalue=60, reset=60

### Change the threshold level to a value below the current value.  
### This should cause the threshold policy to fire an alert.

### Common-thresholds.nmis

      'ssCpuRawUser' => {
        'item' => 'ssCpuRawUser',
        'event' => 'Proactive CPU User',
        'select' => {
          'default' => {
            'value' => {
              'fatal' => '90',
              'critical' => '80',
              'major' => '70',
              'minor' => '65',
              'warning' => '1',
            }
          }
        }
      },
##############################

### In order to test with the new threshold a nmis update must be executed.
### From the bash prompt:

/usr/local/nmis8/bin/nmis.pl type=update

/usr/local/nmis8/bin/nmis/pl type=threshold debug=true

### Now look for the corresponding threshold policy in the debug output.

22:43:59 getThresholdLevel, Start theshold=ssCpuRawUser, index=
22:43:59 getThresholdLevel, found threshold=ssCpuRawUser entry=default
22:43:59 getThresholdLevel, threshold=ssCpuRawUser, item=ssCpuRawUser, value=2.18
22:43:59 getThresholdLevel, result threshold=ssCpuRawUser, level=Warning, value=2.18, thrvalue=1, reset=0
22:43:59 thresholdProcess, Proactive CPU User, Warning, , value=2.18 reset=0

Log into the GUI and notice the alert present in the NMIS event log.

 

Files

Files that require modification:

  • /usr/local/nmis8/models/Model-Some-switch.nmis
  • /usr/local/nmis8/models/Common-database.nmis
  • /usr/local/nmis8/models/Common-header.nmis
  • /usr/local/nmis8/models/Common-stats.nmis
  • /usr/local/mmis8/models/Common-threshold.nmis

Relationship of the files to each other

It may be useful to visualize how the files interact with each other.

Common Attributes

There are several common attributes that must match between these files in order for thresholding to work.  In an attempt to demonstrate the relationship between these variables we'll use the following labels.  Please reference the code block below for where they should reside. 

  • alpha
  • bravo
  • charlie
  • delta
  • echo

alpha - This is the threshold variable set in the Model-Some-switch.nmis file and must be referenced in the name section of the Common-threshold nmis file.

bravo - This the the graphtype variable set in the Model-Some-switch.nmis file.   This sets the name of the graph that NMIS will invoke for a particular RRD file.  In order for the graph heading to be correct this variable with its associated description needs to be added to the Common-heading.nmis file.

charlie - This is the name of the RRD file that the data retrieved via SNMP will be written to.  This variable is set in the Model-Some-switch.nmis file.  It must also be referenced in the Common-database.nmis file in order for the RRD to be created and updated in the correct directory.  This variable is referenced in the Common-stats.nmis file in order for NMIS to know where to find statistical data.

delta - This is specific to an SNMP OID and is set in the Model-Some-switch.nmis file.  This will be the 'DS' (Data Source) title within the RRD file that will contain the actual data returned for a specific OID.  This variable is referenced in the Common-stats.nmis file in order to provide the DS for the specific data that needs to be presented.

echo - This variable is set in the Common-stats.nmis file.  It is used to make computations on the zebra variable in the RRD language.  This variable is then passed to the Common-threshold.nmis file in order to fire and event.

 

  • No labels