check_hadoop_jmx.pl (Advanced Nagios Plugins Collection)

Project Notes

Hari Sekhon - https://github.com/harisekhon/nagios-plugins

check_hadoop_jmx.pl

Nagios Plugin to parse metrics from a given Hadoop daemon's /jmx page

Specify ports depending on which daemon you're trying to get JMX from: HDFS NameNode = 50070 / DataNode = 50075, HBase Master = 60010 / RegionServer = 60030 (for some reason on Hortonworks Sandbox 2.3
the ports for HBase are instead 16010 and 16030)

Make sure you specify --mbean in prod, leave it out with --all-metrics and -vv only for exploring what is available. Nagios has a char limit and will truncate the output, and the perfdata at the end
would be lost.

Tested on Hadoop 2.6 / 2.7 NameNode & DataNode, HBase 0.98 Master & RegionServer on Hortonworks HDP 2.2 & HDP 2.3

usage: check_hadoop_jmx.pl [ options ]

-H --host Hadoop host ($HADOOP_HOST, $HOST)
-P --port Hadoop port ($HADOOP_PORT, $PORT)
-b --bean Bean to check (see --list-beans)
-m --metrics Metric(s) to collect, comma separated. Output in the order specified for convenience. Optional thresholds will only be applied when a single metrics is given
-a --all-metrics Grab all metrics. Useful if you don't know what to monitor yet or just want to graph everything
-w --warning Warning threshold or ran:ge (inclusive)
-c --critical Critical threshold or ran:ge (inclusive)
-e --expected Expected string match when specifying a single field to check. Use this to check non-float fields such as settings eg. -b Hadoop:service=NameNode,name=NameNodeStatus -m State -e
active
--list-beans List all beans returned by HBase jmx page
-t --timeout Timeout in secs (default: 10)
-v --verbose Verbose mode (-v, -vv, -vvv ...)
-h --help Print description and usage options
-V --version Print version and exit