Deep dive into Spring Boot Actuator HTTP metrics

Actuator Metrics

As reported in Michał Bobowski post, we heavily use Spring Boot Actuator metrics system based on Micrometer. It provides a set of practical metrics regarding JVM stats like CPU or memory utilization. Our applications have to meet the most sophisticated needs of our clients thus we try to take advantage of http.server.request endpoint.

Introduction

By default, Spring Boot Actuator gathers endpoint statistics for all classes annotated with @RestController. It registers a WebMvcMetricsFilter bean, which is responsible for timing a request. A special TimingContext attribute is attached to the request so that Spring Boot knows when the request started.

Actuator metrics model

When you call http://localhost:8080/actuator/metrics/http.server.request endpoint you will get something similar to this:

{
  "name": "http.server.requests",
  "description": null,
  "baseUnit": "milliseconds",
  "measurements": [
    {
      "statistic": "COUNT",
      "value": 12
    },
    {
      "statistic": "TOTAL_TIME",
      "value": 21487.256644
    },
    {
      "statistic": "MAX",
      "value": 2731.787888
    }
  ],
  "availableTags": [
    {
      "tag": "exception",
      "values": [
        "None",
        "RuntimeException"
      ]
    },
    {
      "tag": "method",
      "values": [
        "GET"
      ]
    },
    {
      "tag": "uri",
      "values": [
        "/example/success"
      ]
    },
    {
      "tag": "outcome",
      "values": [
        "SERVER_ERROR",
        "SUCCESS"
      ]
    },
    {
      "tag": "status",
      "values": [
        "500",
        "200"
      ]
    }
  ]
}

You will surely see the measurements section. It provides types and values of statistics recorded at a certain point in time. Types of statistics are ones described in Statistics enum.
Another one is the availableTags section, which contains a set of default tags distinguishing each metric by URI, status, or method. You can easily put your tags there like a host or container. If you want to check metric for a particular tag, Actuator lets you do this by using tag query http://localhost:8080/actuator/metrics/http.server.request?tag=status:200

Metric system model

However, each monitoring system has its own metrics model and therefore uses different names for the same things. In our case, we use Influx Registry.
Let’s look into InfluxMeterRegistry class implementation.

private Stream writeTimer(Timer timer) {
    final Stream fields = Stream.of(
        new Field("sum", timer.totalTime(getBaseTimeUnit())),
        new Field("count", timer.count()),
        new Field("mean", timer.mean(getBaseTimeUnit())),
        new Field("upper", timer.max(getBaseTimeUnit()))
    );

    return Stream.of(influxLineProtocol(timer.getId(), "histogram", fields));
}

We see which field in influx corresponds to actuators measurement. Moreover, our registry equips us with an additional mean field, which is basically TOTAL_TIME divided by COUNT. Therefore we don’t need to calculate it manually inside our monitoring system.

Summary

(1) Be aware that the Actuator metric model directly corresponds to Micrometer model
(2) When it comes to timing requests carefully choose the step in which metrics are exported
(3) Do not mix composing metric values with aggregations, selectors, and transformations, e.g. mean(mean)

You May Also Like

OVal – validate your models quickly and effortlessly!

Some time ago one of the projects at work required me to validate some Java POJOs. Theses were my model classes and I've been creating them from incoming WebService requests. One would say that XSD would be sufficient for the task, for parts of this va...Some time ago one of the projects at work required me to validate some Java POJOs. Theses were my model classes and I've been creating them from incoming WebService requests. One would say that XSD would be sufficient for the task, for parts of this va...

Simple trick to DRY your Grails controller

Grails controllers are not very DRY. It's easy to find duplicated code fragments in default generated controller. Take a look at code sample below. It is duplicated four times in show, edit, update and delete actions:

class BookController {
def show() {
def bookInstance = Book.get(params.id)
if (!bookInstance) {
flash.message = message(code: 'default.not.found.message', args: [message(code: 'book.label', default: 'Book'), params.id])
redirect(action: "list")
return
}
[bookInstance: bookInstance]
}
}

Why is it duplicated?

There is a reason for that duplication, though. If you move this snippet to a method, it can redirect to "list" action, but it can't prevent controller from further execution. After you call redirect, response status changes to 302, but after method exits, controller still runs subsequent code.

Solution

At TouK we've implemented a simple trick to resolve that situation:

  1. wrap everything with a simple withStoppingOnRender method,
  2. whenever you want to render or redirect AND stop controller execution - throw EndRenderingException.

We call it Big Return - return from a method and return from a controller at once. Here is how it works:

class BookController {
def show(Long id) {
withStoppingOnRender {
Book bookInstance = Book.get(id)
validateInstanceExists(bookInstance)
[bookInstance: bookInstance]
}
}

protected Object withStoppingOnRender(Closure closure) {
try {
return closure.call()
} catch (EndRenderingException e) {}
}

private void validateInstanceExists(Book instance) {
if (!instance) {
flash.message = message(code: 'default.not.found.message', args: [message(code: 'book.label', default: 'Book'), params.id])
redirect(action: "list")
throw new EndRenderingException()
}
}
}

class EndRenderingException extends RuntimeException {}

Example usage

For simple CRUD controllers, you can use this solution and create some BaseController class for your controllers. We use withStoppingOnRender in every controller so code doesn't look like a spaghetti, we follow DRY principle and code is self-documented. Win-win-win! Here is a more complex example:

class DealerController {
@Transactional
def update() {
withStoppingOnRender {
Dealer dealerInstance = Dealer.get(params.id)
validateInstanceExists(dealerInstance)
validateAccountInExternalService(dealerInstance)
checkIfInstanceWasConcurrentlyModified(dealerInstance, params.version)
dealerInstance.properties = params
saveUpdatedInstance(dealerInstance)
redirectToAfterUpdate(dealerInstance)
}
}
}

Private fields and methods are not private in groovy

I used to code in Java before I met groovy. Like most of you, groovy attracted me with many enhancements. This was to my surprise to discover that method visibility in groovy is handled different than Java!

Consider this example:

class Person {
private String name
public String surname

private Person() {}

private String signature() { "${name?.substring(0, 1)}. $surname" }

public String toString() { "I am $name $surname" }
}

How is this class interpreted with Java?

  1. Person has private constructor that cannot be accessed
  2. Field "name" is private and cannot be accessed
  3. Method signature() is private and cannot be accessed

Let's see how groovy interpretes Person:

public static void main(String[] args) {
def person = new Person() // constructor is private - compilation error in Java
println(person.toString())

person.@name = 'Mike' // access name field directly - compilation error in Java
println(person.toString())

person.name = 'John' // there is a setter generated by groovy
println(person.toString())

person.@surname = 'Foo' // access surname field directly
println(person.toString())

person.surname = 'Bar' // access auto-generated setter
println(person.toString())

println(person.signature()) // call private method - compilation error in Java
}

I was really astonished by its output:

I am null null
I am Mike null
I am John null
I am John Foo
I am John Bar
J. Bar

As you can see, groovy does not follow visibility directives at all! It treats them as non-existing. Code compiles and executes fine. It's contrary to Java. In Java this code has several errors, pointed out in comments.

I've searched a bit on this topic and it seems that this behaviour is known since version 1.1 and there is a bug report on that: http://jira.codehaus.org/browse/GROOVY-1875. It is not resolved even with groovy 2 release. As Tim Yates mentioned in this Stackoverflow question: "It's not clear if it is a bug or by design". Groovy treats visibility keywords as a hint for a programmer.

I need to keep that lesson in mind next time I want to make some field or method private!