Skip to content

go dogstatsd client library for datadog

License

Notifications You must be signed in to change notification settings

blakearnold/datadog-go

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Build Status

Datadog Go

Godoc license

datadog-go is a library that provides a DogStatsD client in Golang.

Go 1.12+ is officially supported. Older versions might work but are not tested.

The following documentation is available:

New major version

The new major version v5 is now the default. All new features will be added to this version and only bugfixes will be backported to v4 (see v4 branch).

v5 introduce a number of breaking changes compare to v4, see the CHANGELOG for more information.

Note that the import paths for v5 and v4 are different:

  • v5: github.com/DataDog/datadog-go/v5/statsd
  • v4: github.com/DataDog/datadog-go/statsd

When migrating to the v5 you will need to upgrade your imports.

Installation

Get the code with:

$ go get github.com/DataDog/datadog-go/v5/statsd

Then create a new DogStatsD client:

package main

import (
    "log"
    "github.com/DataDog/datadog-go/v5/statsd"
)

func main() {
    statsd, err := statsd.New("127.0.0.1:8125")
    if err != nil {
        log.Fatal(err)
    }
}

Find a list of all the available options for your DogStatsD Client in the Datadog-go godoc documentation or in Datadog public DogStatsD documentation.

Supported environment variables

  • If the addr parameter is empty, the client will:
    • First use the DD_DOGSTATSD_URL environment variables to build a target address. This must be a URL that start with either udp:// (to connect using UDP) or with unix:// (to use a Unix Domain Socket). Example for UDP url: DD_DOGSTATSD_URL=udp://localhost:8125 Example for UDS: DD_DOGSTATSD_URL=unix:///var/run/datadog/dsd.socket Example for Windows named pipeDD_AGENT_HOST=\\.\pipe\my_windows_pipe
    • Fallback to the DD_AGENT_HOST environment variables to build a target address. Example: DD_AGENT_HOST=127.0.0.1:8125 for UDP, DD_AGENT_HOST=unix:///path/to/socket for UDS and DD_AGENT_HOST=\\.\pipe\my_windows_pipe for Windows named pipe.
      • If DD_AGENT_HOST has no port it will default the port to 8125
      • You can use DD_AGENT_PORT to set the port if DD_AGENT_HOST does not have a port set for UDP Example: DD_AGENT_HOST=127.0.0.1 and DD_AGENT_PORT=1234 will create a UDP connection to 127.0.0.1:1234.
  • If the DD_ENTITY_ID environment variable is found, its value is injected as a global dd.internal.entity_id tag. The Datadog Agent uses this tag to insert container tags into the metrics.

To enable origin detection and set the DD_ENTITY_ID environment variable, add the following lines to your application manifest:

env:
  - name: DD_ENTITY_ID
    valueFrom:
      fieldRef:
        fieldPath: metadata.uid
  • DD_ENV, DD_SERVICE, and DD_VERSION can be used by the statsd client to set {env, service, version} as global tags for all data emitted.

Unix Domain Sockets Client

Agent v6+ accepts packets through a Unix Socket datagram connection. Details about the advantages of using UDS over UDP are available in the DogStatsD Unix Socket documentation. You can use this protocol by giving a unix:///path/to/dsd.socket address argument to the New constructor.

Usage

In order to use DogStatsD metrics, events, and Service Checks, the Agent must be running and available.

Metrics

After the client is created, you can start sending custom metrics to Datadog. See the dedicated Metric Submission: DogStatsD documentation to see how to submit all supported metric types to Datadog with working code examples:

Metric names must only contain ASCII alphanumerics, underscores, and periods. The client will not replace nor check for invalid characters.

Some options are suppported when submitting metrics, like applying a sample rate to your metrics or tagging your metrics with your custom tags. Find all the available functions to report metrics in the Datadog Go client GoDoc documentation.

Events

After the client is created, you can start sending events to your Datadog Event Stream. See the dedicated Event Submission: DogStatsD documentation to see how to submit an event to your Datadog Event Stream.

Service Checks

After the client is created, you can start sending Service Checks to Datadog. See the dedicated Service Check Submission: DogStatsD documentation to see how to submit a Service Check to Datadog.

Client side aggregation

Starting with version 5.0.0 (and 3.6.0 in beta), the client offers aggregation or value packing on the client side.

This feature aims at reducing both the number of packets sent to the Agent and the packet drops in very high throughput scenarios.

The aggregation window is 2s by default and can be changed through WithAggregationInterval() option. Note that the aggregation window on the Agent side is 10s for DogStatsD metrics. So for example, setting an aggregation window of 3s in the client will produce a spike in your dashboard every 30 second for counts metrics (as the third 10s bucket on the Agent will receive 4 samples from the client).

Aggregation can be disabled using the WithoutClientSideAggregation() option.

The telemetry datadog.dogstatsd.client.metrics is unchanged and represents the number of metrics before aggregation. New metrics datadog.dogstatsd.client.aggregated_context and datadog.dogstatsd.client.aggregated_context_by_type have been introduced. See the Monitoring this client section.

"Basic" aggregation

Enabled by default, the client will aggregate gauge, count and set.

This can be disabled with the WithoutClientSideAggregation() option.

"Extended" aggregation

This feature is only compatible with Agent's version >=6.25.0 && <7.0.0 or Agent's versions >=7.25.0.

Disabled by default, the client can also pack multiple values for histogram, distribution and timing in one message. Real aggregation is not possible for those types since the Agent also aggregates and two aggregation levels would change the final value sent to Datadog.

When this option is enabled, the agent will buffer the metrics by combination of metric name and tags, and send them in the fewest number of messages.

For example, if we sample 3 times the same metric. Instead of sending on the network:

my_distribution_metric:21|d|#all,my,tags
my_distribution_metric:43.2|d|#all,my,tags
my_distribution_metric:1657|d|#all,my,tags

The client will send only one message:

my_distribution_metric:21:43.2:1657|d|#all,my,tags

This will greatly reduce network usage and packet drops but will slightly increase the memory and CPU usage of the client. Looking at the telemetry metrics datadog.dogstatsd.client.metrics_by_type / datadog.dogstatsd.client.aggregated_context_by_type will show the aggregation ratio for each type. This is an interesting data to know how useful extended aggregation is to your app.

This can be enabled with the WithExtendedClientSideAggregation() option.

Maximum samples per context

This feature is best coupled with the previous aggregation mechanism. It allows to limit the number of samples per context for histogram, distribution and timing metrics.

This can be enabled with the WithMaxSamplesPerContext(n int) option. When enabled up to n samples will be kept in per context. The default value is 0 which means no limit.

The selection of the samples is using an algorithm that tries to keep the distribution of kept sample over time uniform.

Performance / Metric drops

Monitoring this client

This client automatically injects telemetry about itself in the DogStatsD stream. Those metrics will not be counted as custom and will not be billed. This feature can be disabled using the WithoutTelemetry option.

See Telemetry documentation to learn more about it.

Tweaking kernel options

In very high throughput environments it is possible to improve performance by changing the values of some kernel options.

Unix Domain Sockets

  • sysctl -w net.unix.max_dgram_qlen=X - Set datagram queue size to X (default value is usually 10).
  • sysctl -w net.core.wmem_max=X - Set the max size of the send buffer for all the host sockets.

Maximum packets size in high-throughput scenarios

In order to have the most efficient use of this library in high-throughput scenarios, default values for the maximum packets size have already been set to have the best usage of the underlying network. However, if you perfectly know your network and you know that a different value for the maximum packets size should be used, you can set it with the option WithMaxBytesPerPayload. Example:

package main

import (
    "log"
    "github.com/DataDog/datadog-go/v5/statsd"
)

func main() {
    statsd, err := statsd.New("127.0.0.1:8125", WithMaxBytesPerPayload(4096))
    if err != nil {
        log.Fatal(err)
    }
}

Development

Run the tests with:

$ go test

License

datadog-go is released under the MIT license.

Credits

Original code by ooyala.

About

go dogstatsd client library for datadog

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Go 100.0%