Bash

This document explains the configuration of SHELL scripts in defining a custom monitor.

SHELL script configuration

Only Bourne Again SHell (bash), Korn shell (ksh) support arrays. Because the parameters need to be parsed and assign to arrays, only bash, ksh shell scripts must be used for writing custom monitor plugins.

Bourne shell (sh) does not support arrays.

The shebang lines #!/bin/bash, #!/bin/sh are required for the scripts. For example:

./sample.sh 
    "/metricName::metricName1|metricName2" 
    "/metric::displayName1|displayName2" 
    "/warn::warn1|warn2" "/critical::crit1|crit2" 
    "/alert::do_alert1|do_alert2" 
    "/params::'args_string1|args_string2'"
#!/bin/bash 
LANG=C`

Parsing arguments

Use the following block of code to parse arguments in your custom monitor bash scripts.

 script_dir=$(dirname $0)
 params=`echo $@ | tr " " "\n"`
 for param in $params
 do
 	case "$param" in
		/metricName::*)
			IFS="|" read -ra metricNames<<<"${param#*::}"
		;;
		/metric::*)
			IFS="|" read -ra metrics<<<"${param#*::}"
		;;
		/warn::*) 
			IFS="|" read -ra warnings<<<"${param#*::}"
		;;
		/critical::*)
			IFS="|" read -ra criticals<<<"${param#*::}"
		;;
		/alert::*)
			IFS="|" read -ra doAlerts<<<"${param#*::}"
		;;
		/params::*)
			userParamsString=$(echo ${param#*::} | sed -e "s/'//g")
			IFS="|" read -ra userParams<<<"$userParamsString"
		;;
		*)
	esac
 done

Accessing variables

The following sample shows how to access the variables.

 arrayLen=${#metricNames[@]}
 for (( x=0; x<$arrayLen; x++ ));
    echo ${metricNames[x]}
     echo ${metrics[x]}
     echo ${warnings[x]}
     echo ${criticals[x]}
     echo ${doAlerts[x]}
     echo ${userParams[x]}
     echo ""
 done

You need metric display names for graphical data in JSON payload. Example: “metric” : “displayName1”

Warning and critical thresholds help to calculate the alert state of metric.

Displaying graphs

For graphing the metric values and for alerting, the script must write the output to the console in the below json format.

{
    "metric" : "displayName1",
    "component" : "instance1",
    "value" : "value1",
    "state" : "alert state1",
    "alert_desc" : "alert description1",
    "hostname/address/id" : "host1"
},
{
    "metric" : "displayName2",
    "component" : " instance2",
    "value" : "value2",
    "state" : "alert state2",
    "alert_desc" : "alert description2",
    "hostname/address/id" : "host2"
}

Provide one of the following values: hostname, address, or id (only when metrics need to post on remote hosts).

• If a user wants to send metric data to a remote host named XYZ, JSON object should be represented as “hostname”: XYZ.
• If a user wants to send metric data to a remote host with IP address x.x.x.x, JSON object should be represented as “address”: x.x.x.x.

Possible values for state are: #OK #Warning #Critical

Viewing graphical representations

To view the graphical representation of the monitoring:

1. From All Clients, select a client.

2. Go to Infrastructure > Resources, select a resource type.

3. Click the resource name.

4. From the left pane, click Metrics. The graph is displayed with display names: displayName1, displayName2.

   • The graph of displayName1 consists of three values val1, val2, val3 that are plotted against key1, key2 and key3 respectively.
   • The graph of displayName2 consists of two values val1, val2 that were plotted against key1 and key2 respectively.