Using Prometheus instrumenters with Gridgo Boot¶

This tutorial will help you get up and running with Gridgo Boot Prometheus instrumenters. It is recommended that you read the instrumenters basic first. What it’ll do:

Create a Prometheus instrumenter to measure a processor throughput and latency.
Expose Prometheus metrics through a HTTP endpoint.
Utilize the metrics in Grafana dashboard.

Installation¶

To start using Prometheus instrumenters, first you need to add it in your pom.xml

<dependency>
    <groupId>io.gridgo</groupId>
    <artifactId>gridgo-extras-prometheus</artifactId>
    <version>${gridgo.version}</version>
</dependency>

Create and register an instrumenter¶

The next step is to create and register the instrumenter in Gridgo Registry. This can be done using a @RegistryInitializer method, which is probably placed inside your initialization class (the one you pass to GridgoApplication.run(…), usually Main class or Initializer)

@RegistryInitializer
public static void initRegistry(Registry registry) {
    // create a Prometheus summary instrumenter, referring to Prometheus documentation for more
    var instrumenter = new PrometheusSummaryTimeInstrumenter("my_summary", "My Summary");
    // register it in our registry
    registry.register("myInstrumenter", instrumenter);
}

Attach the registered instrumenter to the processor¶

Simply add a @Instrumenter to your processor class with name you registered earlier.

@Gateway("my_gateway")
@Connector("${my_connector_endpoint}")
@Instrumenter("myInstrumenter")
public class MyProcessor extends AbstractProcessor {

    @Override
    public void process(RoutingContext rc, GridgoContext gc) {
        // handle the message as usual
    }
}

Expose the metrics via a HTTP endpoint¶

Now everything is setup, but Prometheus is pull-based, so you need to expose metrics via a HTTP endpoint so that the Prometheus server can scrape it. The MetricsProcessor is already provided for your convenience. my_metrics_endpoint might be something like vertx:http://0.0.0.0:8080/metrics

@Gateway("metrics")
@Connector("${my_metrics_endpoint}")
public class MyMetricsProcessor extends MetricsProcessor {

    public MyMetricsProcessor() {
        super("my_metrics");
    }
}

When you run the application, it should be accessible by going to http://localhost:8080/metrics

Visualize the metrics¶

Ask your administrator or infrastructure team to scrape it using the endpoint you have specified earlier. Then you can query it use something like:

rate(my_metrics_my_summary_time_count[30s]): Return the average throughput of last 30 seconds rate(my_metrics_my_summary_time_sum[30s]) / rate(my_metrics_my_summary_time_count[30s]): Return the average latency of last 30 seconds