check_graphite is a nagios module to query graphite

Example

check_graphite -H 'http://my.graphite.host

check_graphite -H "http://your.graphite.host/render" -M collectd.somebox.load.load.midterm -w 1 -c 2 -N load WARNING|load=1.4400000000000002;;;;

check_graphite accepts the following options:

• -H or --endpoint: the graphite HTTP endpoint which can be queried
• -M or --metric: the metric expression which will be queried, it can be an expression
• -F or --from: time frame for which to query metrics, defaults to "30seconds"
• --to: will exclude the time frame's worth of recent data
• --minimum: require a minimum interval for returned data; if actual datapoints don't span interval, do UNKNOWN
• -N or --name: name to give to the metric, defaults to "value"
• -U or --username: username used for basic authentication
• -P or --password: password used for basic authentication
• -w: warning threshold for the metric
• -c: critical threshold for the metric
• -t: timeout after which the metric should be considered unknown
• --ignore-missing: return OK when the metric doesn't exist yet e.g. errors have not occurred
• --projection: Warn on a value linearly extrapolated into the future, defaults to "2days"
• --p-threshold: Return unknown unless p-value exceeds the given value 0-1

How it works

check_graphite, asks for a small window of metrics, and computes an average over the last valid points collected, it then checks the value against supplied thresholds. Thresholds are expressed in the format given in The Nagios Developer Guidelines.

NaN values are not taken into account in the average

When running with --projection, note that you probably want to set a bigger --from window so you get a reasonable projection. Both --from and --projection accepts arguments in Graphite format.