NetApp-probes

Nagios / Graphite probe(s) for monitoring of NetApp filers.

Features

Support for different groups of perf. metrics available through the NetApp SDK
- Currently available: aggregates, volumes, nfsv3, cifs, processor, system, interfaces
- Easy to add new ones
Uses meta-information APIs from NetApp SDK to process metrics auto-magically (i.e. calculate deltas, rates, averages, etc. as appropriate)
Caching of metric descriptions and metric values necessary for calculations (e.g. rates) in JSON files to reduce load on the filers
Supports high frequency collection (currently used on our filers with 5 second intervals)
Used in production for NetApp FAS6240 filers
Support for exploration of the perf. monitoring APIs of the NetApp SDK (e.g. list all available metrics for objects, list all objects)
Works under Linux / Windows (tested with Microsoft Server 2012 with Cygwin)
Optimized NetApp SDK API calls for high performance / low load on the filers
Extensive logging capabilities
Can be used as Nagios probe
- Generate alerts based on performance metrics (ranges, etc.)
- Use multiple metrics in one call to alert on
- Return metrics to generate graphs in Nagios
Graphite mode (compatible with InfluxDB, etc.)
- Collect perf. metrics continuesly and write them into a metrics database of choice via graphite protocol

Limitations

Currently only compatible with NetApp 7-Mode APIs
One probe per filer model

Installation

Get NetApp perl SDK

Download netapp-manageability-sdk-5.4 from http://support.netapp.com/NOW/cgi-bin/software
Extract and copy the lib/perl/NetApp directory with the perl modules to somewhere where it can be found by perl (e.g. directly into the root path of this repo)
Find the API documentation here: ./netapp-manageability-sdk-5.4P1/doc/perldoc/Ontap7ModeAPI.html#perf_object_get_instances

Linux: install necessary perl modules available in most distros as packages

lwp-useragent
xml-parser
Monitoring::Plugin
Log::Log4perl
Net::SSLeay

Linux: install missing perl modules directly from CPAN

For some modules dev packages may need to be installed first

perl -MCPAN -e shell
cpan> install Monitoring::Plugin        (For Nagios stuff)
cpan> install Log::Log4perl             (For logging)
cpan> install JSON                      (For json rendering in caching files)
cpan> install File::Slurp
cpan> install Switch                    (CHORNY/Switch-2.17.tar.gz)
cpan> install Clone                     (GARU/Clone-0.37.tar.gz)
cpan> install Net::Graphite             (v0.16)
cpan> install IO::Async                 (For Timer loop: IO::Async::Timer::Periodic, v0.70)
cpan> install Time::HiRes               (For high resolution (ms) times, v1.9732)

Windows

A list of packages installed for a working Cygwin deployment under Windows Server 2012 is commited into the repository (packagelist.cygwin)

The packages from the list should be able to be installed in Cygwin with the following command (untested):

setup-x86_64 -P `awk 'NR==1{printf \$1}{printf ",%s", \$1}' packagelist.cygwin`

Examples

Get system stats as metrics and write them out via graphite protocol, every 5s:

check_netapp.pl -H <filer-ip> -U <user> -P <password> -o graphite -w 5 -s system

Get nfsv3 and processorr stats and write them out via graphite protocol, every 5s:

check_netapp.pl -H <filer-ip> -U <user> -P <password> -o graphite -w 5 -s nfsv3,processor

A more complex example (graphite, every 5s, all interfaces, two volumes, two aggregates):

./check_netapp.pl -H <filer-ip> -U <user> -P <pwd> -o graphite -w 5 -s cifs,nfsv3,processor,system,interface=all,aggregate=<aggregate1>,aggregate=<aggregate2>,volume=<volume1>,volume=<volume2>