When consuming event streams in Apache Kafka, there are various ways of handling exceptions. This blog post will give a detailed example of publishing dead-letter records with Spring Kafka. Areas where we deviate from the defaults will be highlighted, along with the considerations, and tests are provided.
A Spring Boot application typically consists of several components handling the business functionality and probably some configuration to configure all these components. This configuration consists of defining some properties, setting up some beans with the right conditions and dependencies and wrapping it all together into a class structure. Nevertheless, the configuration of our Spring Boot application is also code. Let’s threat is as code.
In this blog we will se how we can improve our Spring Boot Configuration by splitting up the configuration from the properties and how this effects the design principles.
WebFlux is the reactive web framework for Spring. The programming model is easy but using it could be cumbersome if you don’t know the consequences when used incorrectly. This post will show what the consequences are when the reactive-stack is not used correctly. It will answer the question: why is my (health) endpoint sometimes so slow?
TL DR; Don’t block the event loop
Spring offers several frameworks to implement server side rendered web pages and REST APIS. In this blog I compare three options:
traditional, servlet based (spring-web),
reactive, Netty based (spring-webflux) and
DSL, reactive, Netty based (spring-jafu)
In a previous post we learned that configuration property values can be passed via environment variables. With Spring Boot we can also pass the values using Java system properties. So if we have a property sample.message then we can use -Dsample.message=value to pass a value when we run the application. If we use the Spring Boot Gradle plugin we must reconfigure the bootRun task to pass Java system properties from the command-line.
Let's reuse our sample application from the previous blog post:
Spring Boot has many options for externalising the configuration of our application. One of the options it to use OS environment variables. We need to follow certain rules to name the environment variable and then Spring Boot will use the value of variable to set a configuration property. The property name must be in uppercase and any dots need to be replaced with underscores. For example the configuration property sample.message is set with the environment variable SAMPLE_MESSAGE. This feature can be useful in a continuous integration environment where we can set environment variables or just when developing locally. The nice thing is that this also works when we use the Spring Boot Gradle plugin. The environment variables are passed on to the Java process that the bootRun task starts.
The following source file is a simple Spring Boot command-line application. The sample.message property can be configured as by Spring. If there is no value set the default value "default" is used.
The auto-configuration feature in Spring Boot adds beans to our application context based on conditions. For example based on the availability of a class on the class path or a environment property beans are enabled or disabled. We must apply the @EnableAutoConfiguration or @SpringBootApplicaiton in our Spring Boot application to make this work. To get an overview of all the configurations that had positive and negative conditional matches in our application we can use the --debug command-line option. This will print out a report to System.out with a complete overview. We can check why a configuration is applied or not.
In the following Gradle build file we add the option --debug to the args property of the bootRun task:
When we develop a Spring Boot application we can hot reload newly compiled classes using Gradle. The bootRun task of the Spring Boot Gradle plugin has support for Spring Loaded. We must add a dependency to Spring Loaded as a dependency for the classpath configuration in the buildscript block. If we now use the bootRun task everything is ready to reload changed classes. Now we can use the continuous build feature of Gradle to listen for changes in our source files to recompile them. The recompiled classes are then reloaded in the running Spring Boot application started with bootRun. We start a second Gradle instance with the -t option for the classes task. This way when we change a source file it gets recompiled automatically and reloaded.
The following build script shows how we add Spring Loaded:
In Spring we can use the @Value annotation to set property or arguments values based on a SpEL expression. If we want to use the @Value annotation for a constructor argument we must not forget to add the @Autowired annotation on the constructor as well.
// File: sample/Message.groovy
package sample
import org.springframework.beans.factory.annotation.*
import org.springframework.stereotype.*
@Component
class Message {
final String text
// Use @Autowired to get @Value to work.
@Autowired
Message(
// Refer to configuration property
// app.message.text to set value for
// constructor argument text.
@Value('${app.message.text}') final String text) {
this.text = text
}
}
