Observability

go-redis has built-in support for OpenTelemetry (OTel) instrumentation to collect metrics. This can be very helpful for diagnosing problems and improving the performance and connection resiliency of your application. See the [Observability overview]({{< relref "/develop/clients/observability" >}}) for an introduction to Redis client observability and a reference guide for the available metrics.

This page explains how to enable and use OTel instrumentation in go-redis using an example configuration for a local Grafana instance. See our observability demonstration repository on GitHub to learn how to set up a suitable Grafana dashboard.

Installation

Install OTel support for go-redis with the following commands:

go get github.com/redis/go-redis/extra/redisotel-native/v9
go get go.opentelemetry.io/otel

Import

Start by importing the required OTel and Redis modules:

{{< clients-example set="observability" step="import" lang_filter="Go" description="Foundational: Import required libraries for Redis observability, OpenTelemetry metrics, and Redis operations" difficulty="beginner" >}} {{< /clients-example >}}

Configure the meter provider

Otel uses a Meter provider to create the objects that collect the metric information. The example below configures a meter provider to export metrics to a local Otel collector every 10 seconds, but see the OpenTelemetry Go docs to learn more about other export options.

{{< clients-example set="observability" step="setup_meter_provider" lang_filter="Go" description="Foundational: Configure a meter provider to export metrics to a local Grafana instance every 10 seconds" difficulty="beginner" >}} {{< /clients-example >}}

Configure the Redis client

You configure the client library for OTel only once per application. This will enable OTel for all Redis connections you create. The example below shows the options you can pass to the observability instance via the redisotel.Config object during initialization.

{{< clients-example set="observability" step="client_config" lang_filter="Go" description="Foundational: Configure Redis observability with a list of metric groups and optional command filtering and privacy controls" difficulty="beginner" >}} {{< /clients-example >}}

The available option methods for redisotel.Config are described in the table below.

Custom histogram buckets

For the histogram metrics, a reasonable default set of buckets is defined, but you can customize the bucket boundaries to suit your needs (the buckets are the ranges of data values counted for each bar of the histogram). Pass an increasing list of float values to the WithHistogramBuckets option when you create the redisotel.Config object. The first and last values in the list are the lower and upper bounds of the histogram, respectively, and the values in between define the bucket boundaries.

Use Redis

Once you have configured the client, all Redis connections you create will be automatically instrumented and the collected metrics will be exported to your configured destination.

The example below shows the simplest Redis connection and a few commands, but see the observability demonstration repository for an example that calls a variety of commands in a more realistic way.

{{< clients-example set="observability" step="use_redis" lang_filter="Go" description="Foundational: Use Redis with automatic instrumentation" difficulty="beginner" >}} {{< /clients-example >}}

Shutdown

When your application exits, you should close the Redis connection and then you should call the Shutdown() method on the ObservabilityInstance and MeterProvider instances to ensure that all pending metrics are exported. You may find it useful to put the shutdown code in a defer statement to ensure that it is called even if the main function exits early due to an error.

{{< clients-example set="observability" step="shutdown" lang_filter="Go" description="Foundational: Shutdown Redis observability" difficulty="beginner" >}} {{< /clients-example >}}