Send metrics to a OpenTSDB server.

OpenTSDB is a distributed, scalable Time Series Database (TSDB) written on top of HBase. OpenTSDB was written to address a common need: store, index and serve metrics collected from computer systems (network gear, operating systems, applications) at a large scale, and make this data easily accessible and graphable.

Thanks to HBase's scalability, OpenTSDB allows you to collect many thousands of metrics from thousands of hosts and applications, at a high rate (every few seconds). OpenTSDB will never delete or downsample data and can easily store billions of data points. As a matter of fact, StumbleUpon uses it to keep track of hundred of thousands of time series and collects over 1 billion data points per day in their main production datacenter.

Imagine having the ability to quickly plot a graph showing the number of DELETE statements going to your MySQL database along with the number of slow queries and temporary files created, and correlate this with the 99th percentile of your service's latency. OpenTSDB makes generating such graphs on the fly a trivial operation, while manipulating millions of data point for very fine grained, real-time monitoring.

The handler per default adds a tag 'hostname' with the hostname where the collection took place. You can add as many as you like via the configuration. The 'tags' config element allows for both comma-separated or space separated key value pairs.

The collectors specify the metrics with a single axis of navigation and often include aggregation counters that OpenTSDB could calculate directly. The openTSDB tagging system supports multiple axis and is better suited for dynamical values. You can read here how OpenTSDB suggests to sepearate tags from metrics. when doing this you also want to remove these aggregate values since they would lead to double counting when requesting OpenTSDB to do the aggregation based on the tags.

The handler currently provides the ability to extract the tags from a metric (and remove aggregate values) for following collectors : CPUCollector HaProxyCollector * MattermostCollector

==== Notes

We don't automatically make the metrics via mkmetric, so either set

set tsd.core.auto_create_metrics = true

in your OpenTSDB configuration or you can run with the null handler and log the output and extract the key values to mkmetric yourself

  • enable it in diamond.conf :

handlers = diamond.handler.tsdb.TSDBHandler


Setting Default Description Type
format {Collector}.{Metric} {timestamp} {value} hostname={host}{tags} str
get_default_config_help get_default_config_help
host str
port 1234 int
server_error_interval 120 How frequently to send repeated server errors int
tags Tags to be added to each metric. They should be key/value pairs. Example : tags = environment=test,datacenter=north str
timeout 5 int
cleanMetrics True Extract tag values from known collectors and make the metrics more OpenTSDB style bool
skipAggregates True Only has effect when cleanMetrics is true. Then the metrics that are considered aggregates are removed bool