check_hadoop_metrics

Project Notes

Checks the /metrics page of a Hadoop NameNode for a given performance metric. Assumes you have exposed these metrics by uncommenting the requisite lines from $HADOOP_CONF/hadoop-metrics.properties and restarting your cluster.

./check_hadoop_metrics.sh --help

Usage: ./check_hadoop_metrics.sh [-H hostname] [-p port] [-C counter] [-b bias:gt|lt|na] [-w warn] [-c crit]
[-H | --hostname] Hostname of the NameNode.
[-p | --port] Port number of the NameNode.
[-C | --counter] Metric to monitor.
[-b | --bias] Evaluation bias: Greater-than (gt), Less-than (lt), Informative (na).
[-w | --warn] Warning threshold.
[-c | --crit] Critical threshold.

commands.cfg:

# Check Hadoop metrics via HTTP.
define command{
command_name check_hadoop_metrics
command_line $USER1$/check_hadoop_metrics.sh -H $HOSTALIAS$ -p 50070 -C $ARG1$ -b $ARG2$ -w $ARG3$ -c $ARG4$
}

Nothing fancy, just some processing on 'curl' command output. Designed with Hadoop in mind.