Nagios check for Pure Storage arrays

Nagios (https://www.nagios.com/) is free monitoring package. This document covers a method of using Nagios to monitor Pure arrays.

SNMP

SNMP is a common method for monitoring systems in a datacenter.

Currently, Pure Storage offers a static (non-trap) SNMP implementation that lists 6 array-wide metrics as part of their MIB:

Viewing: (1.3.6.1.4.1.40482)

PURESTORAGE-MIB::pureArrayReadBandwidth.0 = INTEGER: 0 B/s
PURESTORAGE-MIB::pureArrayWriteBandwidth.0 = INTEGER: 0 B/s
PURESTORAGE-MIB::pureArrayReadIOPS.0 = INTEGER: 0 op/s
PURESTORAGE-MIB::pureArrayWriteIOPS.0 = INTEGER: 0 op/s
PURESTORAGE-MIB::pureArrayReadLatency.0 = INTEGER: 0 us/op
PURESTORAGE-MIB::pureArrayWriteLatency.0 = INTEGER: 0 us/op

This is insufficient for most common monitoring concerns.

REST

A full REST API is available for automation and monitoring Pure Storage arrays.

Consequently, the preferred method for implementing monitoring is via this API.

Configuring the Pure Array

The assumption below is you have already followed instructions for generating an API token. Generating a token is the only array-side configuration required.

Configuring Nagios

This method has nagios execute a local "check" script on the server that queries the Array over the REST API. This is a common design for checks in nagios.

Here is the example Nagios command definition:

# Check PureArray - custom script to check volumes on Pure Arrays
define command {
        command_name check_purearray
        command_line /usr/local/bin/check_purearray.pl $HOSTADDRESS$
}

Here is an example of a service definition applying this check to "my-pure-array1":

define service {
    use                     generic-hourly-service
    host_name               my-pure-array1.company.com
    service_description     Pure Array Usage
    notification_interval   480
    check_command           check_purearray
}

The Perl script used for the check is attached.

To use it, will need to edit it and put the API tokens for each array in the hash, keyed by hostname, at the top as they are in the script. (lines 15-24)

Also of interest: Lines 27 through 31 set the alert thresholds. They are as follows:

27) If "System" on the array is above 10%, critical alert

28 & 29) If the array is above 85% full, warning. If above 90% full, critical.

30 & 31) If a volume is above 85% full, warning. If above 90% full, critical.