StatsdCsharpClient

The simple statsd client for .Net is a robust, easy-to-use way of feeding metrics into a statsd-compatible server. You can use it to log counts, timings and gauges. Features include: * Log counts, timings, gauges, sets and raw metrics * Has an additional API that uses dynamics to create and submit stats * Fault-tolerant client that can be configured to fail silently (with a warning) if misconfigured * IStatsdClient interface for easy mocking in unit tests * Allows for customisation of every output stat to do things like screen metrics before sending * Supports a user-defined prefix to prepend to every metric * Outputs to UDP or TCP .Net 4.5 Users: * Use the StatsdExtensions to define metrics without having to manipulate strings * The dynamic stats builder interface provides a cleaner alternative to creating and logging metrics Licence: MIT


Keywords
etsy, graphite, metrics, statsd, statsd.net
License
MIT
Install
Install-Package StatsdCsharpClient -Version 1.5.0

Documentation

statsd-csharp-client

Build status

A simple c# client library for statsd.net and statsd.

Features

  • Log counts, timings, gauges, sets, calendargrams and raw metrics
  • Has an additional API that uses dynamics to create and submit stats
  • Fault-tolerant client that can be configured to fail silently (with a warning) if misconfigured
  • IStatsdClient interface for easy mocking in unit tests
  • Allows for customisation of every output stat to do things like screen metrics before sending
  • Supports a user-defined prefix to prepend to every metric
  • Send metrics over a UDP or TCP connection

Coming soon:

  • Support for count sampling and histograms
  • batch-and-pump - collecting stats and sending them out in a batch at regular intervals
  • Output to an HTTP endpoint

Download and Install

Install the StatsdCsharpClient via nuget:

PM> Install-Package StatsdCsharpClient

Quickstart

Assuming your server is running on localhost and listening on port 12000:

using StatsdClient;
...
var statsd = new Statsd("localhost", 12000);
// Log a count
statsd.LogCount( "site.hits" );
// Log a gauge
statsd.LogGauge( "site.activeUsers", numActiveUsers );
// Log a timing
statsd.LogTiming( "site.pageLoad", 100 /* milliseconds */ );
// Log a raw metric
statsd.LogRaw ("already.aggregated", 982, 1885837485 /* epoch timestamp */ );
// Log a calendargram
statsd.LogCalendargram("order.completed", "user_13143", CalendargramRetentionPeriod.HOUR);

You can also wrap your code in a using block to measure the latency by using the LogTiming(string) extension method:

using StatsdClient;
...
using (statsd.LogTiming( "site.db.fetchReport" ))
{
  // do some work
}
// At this point your latency has been sent to the server

Dynamic Stats Builder

There's also a nifty set of extension methods that let you define your stats without using strings. Using the example provided above, but now using the builder:

var statsd = new StatsdClient("localhost", 12000);
// Log a count
statsd.count.site.hits += 1;
// Log a gauge
statsd.gauge.site.activeUsers += numActiveUsers;
// Log a timing
statsd.site.pageLoad += 100; /* milliseconds */

TCP Output Channel

Metrics can be delivered over a TCP connection by specifying ConnectionType.Tcp during construction:

var statsd = new Statsd("localhost", 12001);
// Continue as normal

The connection will attempt to reconnect if something goes wrong, and will try three times before giving up. Use the retryOnDisconnect parameter to enable/disable this, and the retryAttempts parameter to specify the number of times to try the request again.

Project Information

Target Runtimes

  • .Net 4.5

Authors

Luke Venediger - lukev@lukev.net and @lukevenediger

See Also