Collect metrics using Micrometer
Create an application that uses the Micrometer metrics library to collect runtime, extension and application metrics and expose them as a Prometheus (OpenMetrics) endpoint.
Pré-requisitos
Para concluir este guia, você precisa:
-
Cerca de 15 minutos
-
Um IDE
-
JDK 17+ installed with
JAVA_HOMEconfigured appropriately -
Apache Maven 3.9.9
-
Opcionalmente, o Quarkus CLI se você quiser usá-lo
-
Opcionalmente, Mandrel ou GraalVM instalado e configurado apropriadamente se você quiser criar um executável nativo (ou Docker se você usar uma compilação de contêiner nativo)
Solução
We recommend that you follow the instructions to create the application step by step, but you can skip right to the solution if you prefer. Either:
-
Clone the git repository:
git clone https://github.com/quarkusio/quarkus-quickstarts.git, or -
Download an archive.
The solution is located in the micrometer-quickstart directory.
1. Crie o projeto Maven
Create a new project with the following command:
Para usuários do Windows:
-
Se estiver usando cmd, (não use barra invertida '\' e coloque tudo na mesma linha)
-
Se estiver usando o Powershell, envolva os parâmetros '-D' entre aspas duplas, por exemplo, '"-DprojectArtifactId=micrometer-quickstart"'
This command generates a Maven project, that imports the micrometer-registry-prometheus extension as a dependency.
This extension will load the core micrometer extension as well as additional library dependencies required to support prometheus.
| To maintain backwards compatibility, the extension uses the Prometheus client v0.x from Micrometer 1.13+, instead of their default v1.x client. |
2. Create a REST endpoint
Let’s first add a simple endpoint that calculates prime numbers.
package org.acme.micrometer;
import java.util.LinkedList;
import java.util.NoSuchElementException;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.PathParam;
import jakarta.ws.rs.Produces;
import io.micrometer.core.instrument.MeterRegistry;
import io.micrometer.core.instrument.Tags;
import io.micrometer.core.instrument.Timer;
@Path("/example")
@Produces("text/plain")
public class ExampleResource {
@GET
@Path("prime/{number}")
public String checkIfPrime(@PathParam("number") long number) {
if (number < 1) {
return "Only natural numbers can be prime numbers.";
}
if (number == 1) {
return number + " is not prime.";
}
if (number == 2 || number % 2 == 0) {
return number + " is not prime.";
}
if (testPrimeNumber(number)) {
return number + " is prime.";
} else {
return number + " is not prime.";
}
}
protected boolean testPrimeNumber(long number) {
for (int i = 3; i < Math.floor(Math.sqrt(number)) + 1; i = i + 2) {
if (number % i == 0) {
return false;
}
}
return true;
}
}Start your application in dev mode:
quarkus dev./mvnw quarkus:dev./gradlew --console=plain quarkusDev2.1. Review automatically generated metrics
The Micrometer extension automatically times HTTP server requests.
Let’s use curl (or a browser) to visit our endpoint a few times:
curl http://localhost:8080/example/prime/256
curl http://localhost:8080/example/prime/7919The Micrometer Prometheus MeterRegistry extension creates an endpoint we can use to observe collected metrics. Let’s take a look at the metrics that have been collected:
curl http://localhost:8080/q/metricsLook for http_server_requests_seconds_count, http_server_requests_seconds_sum, and
http_server_requests_seconds_max in the output.
Dimensional labels are added for the request uri, the HTTP method (GET, POST, etc.), the status code (200, 302, 404, etc.), and a more general outcome field. You should find something like this:
# HELP http_server_requests_seconds
# TYPE http_server_requests_seconds summary
http_server_requests_seconds_count{method="GET",outcome="SUCCESS",status="200",uri="/example/prime/{number}"} 2.0
http_server_requests_seconds_sum{method="GET",outcome="SUCCESS",status="200",uri="/example/prime/{number}"} 0.017385896
# HELP http_server_requests_seconds_max
# TYPE http_server_requests_seconds_max gauge
http_server_requests_seconds_max{method="GET",outcome="SUCCESS",status="200",uri="/example/prime/{number}"} 0.017385896
#| Metrics appear lazily, you often won’t see any data for your endpoint until it is accessed. |
By default, the metrics are exported using the Prometheus format application/openmetrics-text,
you can revert to the former format by specifying the Accept request header to text/plain (curl -H "Accept: text/plain" localhost:8080/q/metrics/).
3. Inject the MeterRegistry
To register meters, you need a reference to the MeterRegistry that is configured and maintained by the Micrometer extension.
The MeterRegistry can be injected into your application as follows:
private final MeterRegistry registry;
ExampleResource(MeterRegistry registry) {
this.registry = registry;
}4. Add a Counter
Counters are used to measure values that only increase.
Let’s add a counter that tracks how often we test a number to see if it is prime. We’ll add a dimensional label (also called an attribute or a tag) that will allow us to aggregate this counter value in different ways.
@GET
@Path("prime/{number}")
public String checkIfPrime(@PathParam("number") long number) {
if (number < 1) {
registry.counter("example.prime.number", "type", "not-natural") (1)
.increment(); (2)
return "Only natural numbers can be prime numbers.";
}
if (number == 1) {
registry.counter("example.prime.number", "type", "one") (1)
.increment(); (2)
return number + " is not prime.";
}
if (number == 2 || number % 2 == 0) {
registry.counter("example.prime.number", "type", "even") (1)
.increment(); (2)
return number + " is not prime.";
}
if (testPrimeNumber(number)) {
registry.counter("example.prime.number", "type", "prime") (1)
.increment(); (2)
return number + " is prime.";
} else {
registry.counter("example.prime.number", "type", "not-prime") (1)
.increment(); (2)
return number + " is not prime.";
}
}| 1 | Find or create a counter called example.prime.number that has a type label with the specified value. |
| 2 | Increment that counter. |
4.1. Review collected metrics
If you did not leave Quarkus running in dev mode, start it again:
quarkus dev./mvnw quarkus:dev./gradlew --console=plain quarkusDevTry the following sequence and look for example_prime_number_total in the plain text
output.
Note that the _total suffix is added when Micrometer applies Prometheus naming conventions to
example.prime.number, the originally specified counter name.
curl http://localhost:8080/example/prime/-1
curl http://localhost:8080/example/prime/0
curl http://localhost:8080/example/prime/1
curl http://localhost:8080/example/prime/2
curl http://localhost:8080/example/prime/3
curl http://localhost:8080/example/prime/15
curl http://localhost:8080/q/metricsNotice that there is one measured value for each unique combination of example_prime_number_total and type value.
Looking at the dimensional data produced by this counter, you can count:
-
how often a negative number was checked:
type="not-natural" -
how often the number one was checked:
type="one" -
how often an even number was checked:
type="even" -
how often a prime number was checked:
type="prime" -
how often a non-prime number was checked:
type="not-prime"
You can also count how often a number was checked (generally) by aggregating all of these values together.
5. Add a Timer
Timers are a specialized abstraction for measuring duration. Let’s add a timer to measure how long it takes to determine if a number is prime.
@GET
@Path("prime/{number}")
public String checkIfPrime(@PathParam("number") long number) {
if (number < 1) {
registry.counter("example.prime.number", "type", "not-natural") (1)
.increment(); (2)
return "Only natural numbers can be prime numbers.";
}
if (number == 1) {
registry.counter("example.prime.number", "type", "one") (1)
.increment(); (2)
return number + " is not prime.";
}
if (number == 2 || number % 2 == 0) {
registry.counter("example.prime.number", "type", "even") (1)
.increment(); (2)
return number + " is not prime.";
}
if (timedTestPrimeNumber(number)) { (3)
registry.counter("example.prime.number", "type", "prime") (1)
.increment(); (2)
return number + " is prime.";
} else {
registry.counter("example.prime.number", "type", "not-prime") (1)
.increment(); (2)
return number + " is not prime.";
}
}
protected boolean timedTestPrimeNumber(long number) {
Timer.Sample sample = Timer.start(registry); (4)
boolean result = testPrimeNumber(number); (5)
sample.stop(registry.timer("example.prime.number.test", "prime", result + "")); (6)
return result;
}| 1 | Find or create a counter called example.prime.number that has a type label with the specified value. |
| 2 | Increment that counter. |
| 3 | Call a method that wraps the original testPrimeNumber method. |
| 4 | Create a Timer.Sample that tracks the start time |
| 5 | Call the method to be timed and store the boolean result |
| 6 | Find or create a Timer using the specified id and a prime label with the result value, and record the duration captured by the Timer.Sample. |
5.1. Review collected metrics
If you did not leave Quarkus running in dev mode, start it again:
quarkus dev./mvnw quarkus:dev./gradlew --console=plain quarkusDevMicrometer will apply Prometheus conventions when emitting metrics for this timer. Specifically, measured durations are converted into seconds and this unit is included in the metric name.
Try the following sequence and look for the following entries in the plain text output:
-
example_prime_number_test_seconds_count— how many times the method was called -
example_prime_number_test_seconds_sum— the total duration of all method calls -
example_prime_number_test_seconds_max— the maximum observed duration within a decaying interval. This value will return to 0 if the method is not invoked frequently.
curl http://localhost:8080/example/prime/256
curl http://localhost:8080/q/metrics
curl http://localhost:8080/example/prime/7919
curl http://localhost:8080/q/metricsLooking at the dimensional data produced by this counter, you can use the sum and the count to calculate how long (on average) it takes to determine if a number is prime. Using the dimensional label, you might be able to understand if there is a significant difference in duration for numbers that are prime when compared with numbers that are not.
6. Add a Gauge
Gauges measure a value that can increase or decrease over time, like the speedometer on a car. The value of a gauge is not accumulated, it is observed at collection time. Use a gauge to observe the size of a collection, or the value returned from a function.
private final LinkedList<Long> list = new LinkedList<>(); (1)
ExampleResource(MeterRegistry registry) {
this.registry = registry;
registry.gaugeCollectionSize("example.list.size", Tags.empty(), list); (2)
}
@GET
@Path("gauge/{number}")
public Long checkListSize(@PathParam("number") long number) { (3)
if (number == 2 || number % 2 == 0) {
// add even numbers to the list
list.add(number);
} else {
// remove items from the list for odd numbers
try {
number = list.removeFirst();
} catch (NoSuchElementException nse) {
number = 0;
}
}
return number;
}| 1 | Define list that will hold arbitrary numbers. |
| 2 | Register a gauge that will track the size of the list. |
| 3 | Create a REST endpoint to populate the list. Even numbers are added to the list, and odd numbers remove an element from the list. |
6.1. Review collected metrics
If you did not leave Quarkus running in dev mode, start it again:
quarkus dev./mvnw quarkus:dev./gradlew --console=plain quarkusDevThen try the following sequence and look for example_list_size in the plain text output:
curl http://localhost:8080/example/gauge/1
curl http://localhost:8080/example/gauge/2
curl http://localhost:8080/example/gauge/4
curl http://localhost:8080/q/metrics
curl http://localhost:8080/example/gauge/6
curl http://localhost:8080/example/gauge/5
curl http://localhost:8080/example/gauge/7
curl http://localhost:8080/q/metricsSummary
Congratulations!
You have created a project that uses the Micrometer and Prometheus Meter Registry extensions to collect metrics.
You’ve observed some of the metrics that Quarkus captures automatically, and have added a Counter and Timer that are unique to the application.
You’ve also added dimensional labels to metrics, and have observed how those labels shape the data emitted by the prometheus endpoint.
