This project provides Zookeeper integrations for Spring Boot applications through autoconfiguration and binding to the Spring Environment and other Spring programming model idioms. With a few annotations, you can quickly enable and configure the common patterns inside your application and build large distributed systems with Zookeeper based components. The provided patterns include Service Discovery and Configuration. The project also provides client-side load-balancing via integration with Spring Cloud LoadBalancer.

1. Install Zookeeper

See the installation documentation for instructions on how to install Zookeeper.

Spring Cloud Zookeeper uses Apache Curator behind the scenes. While Zookeeper 3.5.x is still considered "beta" by the Zookeeper development team, the reality is that it is used in production by many users. However, Zookeeper 3.4.x is also used in production. Prior to Apache Curator 4.0, both versions of Zookeeper were supported via two versions of Apache Curator. Starting with Curator 4.0 both versions of Zookeeper are supported via the same Curator libraries.

In case you are integrating with version 3.4 you need to change the Zookeeper dependency that comes shipped with curator, and thus spring-cloud-zookeeper. To do so simply exclude that dependency and add the 3.4.x version like shown below.

maven
<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>spring-cloud-starter-zookeeper-all</artifactId>
    <exclusions>
        <exclusion>
            <groupId>org.apache.zookeeper</groupId>
            <artifactId>zookeeper</artifactId>
        </exclusion>
    </exclusions>
</dependency>
<dependency>
    <groupId>org.apache.zookeeper</groupId>
    <artifactId>zookeeper</artifactId>
    <version>3.4.12</version>
    <exclusions>
        <exclusion>
            <groupId>org.slf4j</groupId>
            <artifactId>slf4j-log4j12</artifactId>
        </exclusion>
    </exclusions>
</dependency>
gradle
compile('org.springframework.cloud:spring-cloud-starter-zookeeper-all') {
  exclude group: 'org.apache.zookeeper', module: 'zookeeper'
}
compile('org.apache.zookeeper:zookeeper:3.4.12') {
  exclude group: 'org.slf4j', module: 'slf4j-log4j12'
}

2. Service Discovery with Zookeeper

Service Discovery is one of the key tenets of a microservice based architecture. Trying to hand-configure each client or some form of convention can be difficult to do and can be brittle. Curator(A Java library for Zookeeper) provides Service Discovery through a Service Discovery Extension. Spring Cloud Zookeeper uses this extension for service registration and discovery.

2.1. Activating

Including a dependency on org.springframework.cloud:spring-cloud-starter-zookeeper-discovery enables autoconfiguration that sets up Spring Cloud Zookeeper Discovery.

For web functionality, you still need to include org.springframework.boot:spring-boot-starter-web.
When working with version 3.4 of Zookeeper you need to change the way you include the dependency as described here.

2.2. Registering with Zookeeper

When a client registers with Zookeeper, it provides metadata (such as host and port, ID, and name) about itself.

The following example shows a Zookeeper client:

@SpringBootApplication
@RestController
public class Application {

    @RequestMapping("/")
    public String home() {
        return "Hello world";
    }

    public static void main(String[] args) {
        new SpringApplicationBuilder(Application.class).web(true).run(args);
    }

}
The preceding example is a normal Spring Boot application.

If Zookeeper is located somewhere other than localhost:2181, the configuration must provide the location of the server, as shown in the following example:

application.yml
spring:
  cloud:
    zookeeper:
      connect-string: localhost:2181
If you use Spring Cloud Zookeeper Config, the values shown in the preceding example need to be in bootstrap.yml instead of application.yml.

The default service name, instance ID, and port (taken from the Environment) are ${spring.application.name}, the Spring Context ID, and ${server.port}, respectively.

Having spring-cloud-starter-zookeeper-discovery on the classpath makes the app into both a Zookeeper “service” (that is, it registers itself) and a “client” (that is, it can query Zookeeper to locate other services).

If you would like to disable the Zookeeper Discovery Client, you can set spring.cloud.zookeeper.discovery.enabled to false.

2.3. Using the DiscoveryClient

Spring Cloud has support for Feign (a REST client builder), Spring RestTemplate and Spring WebFlux, using logical service names instead of physical URLs.

You can also use the org.springframework.cloud.client.discovery.DiscoveryClient, which provides a simple API for discovery clients that is not specific to Netflix, as shown in the following example:

@Autowired
private DiscoveryClient discoveryClient;

public String serviceUrl() {
    List<ServiceInstance> list = discoveryClient.getInstances("STORES");
    if (list != null && list.size() > 0 ) {
        return list.get(0).getUri().toString();
    }
    return null;
}

3. Using Spring Cloud Zookeeper with Spring Cloud Netflix Components

Spring Cloud Netflix supplies useful tools that work regardless of which DiscoveryClient implementation you use. Feign and Spring Cloud LoadBalancer also work with Spring Cloud Zookeeper.

3.1. Spring Cloud LoadBalancer with Zookeeper

Spring Cloud Zookeeper provides an implementation of Spring Cloud LoadBalancer ServiceInstanceListSupplier. When you use the spring-cloud-starter-zookeeper-discovery, Spring Cloud LoadBalancer is autoconfigured to use the ZookeeperServiceInstanceListSupplier by default.

4. Spring Cloud Zookeeper and Service Registry

Spring Cloud Zookeeper implements the ServiceRegistry interface, letting developers register arbitrary services in a programmatic way.

The ServiceInstanceRegistration class offers a builder() method to create a Registration object that can be used by the ServiceRegistry, as shown in the following example:

@Autowired
private ZookeeperServiceRegistry serviceRegistry;

public void registerThings() {
    ZookeeperRegistration registration = ServiceInstanceRegistration.builder()
            .defaultUriSpec()
            .address("anyUrl")
            .port(10)
            .name("/a/b/c/d/anotherservice")
            .build();
    this.serviceRegistry.register(registration);
}

4.1. Instance Status

Netflix Eureka supports having instances that are OUT_OF_SERVICE registered with the server. These instances are not returned as active service instances. This is useful for behaviors such as blue/green deployments. (Note that the Curator Service Discovery recipe does not support this behavior.) Taking advantage of the flexible payload has let Spring Cloud Zookeeper implement OUT_OF_SERVICE by updating some specific metadata and then filtering on that metadata in the Spring Cloud LoadBalancer ZookeeperServiceInstanceListSupplier. The ZookeeperServiceInstanceListSupplier filters out all non-null instance statuses that do not equal UP. If the instance status field is empty, it is considered to be UP for backwards compatibility. To change the status of an instance, make a POST with OUT_OF_SERVICE to the ServiceRegistry instance status actuator endpoint, as shown in the following example:

$ http POST http://localhost:8081/service-registry status=OUT_OF_SERVICE
The preceding example uses the http command from httpie.org.

5. Zookeeper Dependencies

The following topics cover how to work with Spring Cloud Zookeeper dependencies:

5.1. Using the Zookeeper Dependencies

Spring Cloud Zookeeper gives you a possibility to provide dependencies of your application as properties. As dependencies, you can understand other applications that are registered in Zookeeper and which you would like to call through Feign (a REST client builder), Spring RestTemplate and Spring WebFlux.

You can also use the Zookeeper Dependency Watchers functionality to control and monitor the state of your dependencies.

5.2. Activating Zookeeper Dependencies

Including a dependency on org.springframework.cloud:spring-cloud-starter-zookeeper-discovery enables autoconfiguration that sets up Spring Cloud Zookeeper Dependencies. Even if you provide the dependencies in your properties, you can turn off the dependencies. To do so, set the spring.cloud.zookeeper.dependency.enabled property to false (it defaults to true).

5.3. Setting up Zookeeper Dependencies

Consider the following example of dependency representation:

application.yml
spring.application.name: yourServiceName
spring.cloud.zookeeper:
  dependencies:
    newsletter:
      path: /path/where/newsletter/has/registered/in/zookeeper
      loadBalancerType: ROUND_ROBIN
      contentTypeTemplate: application/vnd.newsletter.$version+json
      version: v1
      headers:
        header1:
            - value1
        header2:
            - value2
      required: false
      stubs: org.springframework:foo:stubs
    mailing:
      path: /path/where/mailing/has/registered/in/zookeeper
      loadBalancerType: ROUND_ROBIN
      contentTypeTemplate: application/vnd.mailing.$version+json
      version: v1
      required: true

The next few sections go through each part of the dependency one by one. The root property name is spring.cloud.zookeeper.dependencies.

5.3.1. Aliases

Below the root property you have to represent each dependency as an alias. This is due to the constraints of Spring Cloud LoadBalancer, which requires that the application ID be placed in the URL. Consequently, you cannot pass any complex path, suchas /myApp/myRoute/name). The alias is the name you use instead of the serviceId for DiscoveryClient, Feign, or RestTemplate.

In the previous examples, the aliases are newsletter and mailing. The following example shows Feign usage with a newsletter alias:

@FeignClient("newsletter")
public interface NewsletterService {
        @RequestMapping(method = RequestMethod.GET, value = "/newsletter")
        String getNewsletters();
}

5.3.2. Path

The path is represented by the path YAML property and is the path under which the dependency is registered under Zookeeper. As described in the previous section, Spring Cloud LoadBalancer operates on URLs. As a result, this path is not compliant with its requirement. That is why Spring Cloud Zookeeper maps the alias to the proper path.

5.3.3. Load Balancer Type

The load balancer type is represented by loadBalancerType YAML property.

If you know what kind of load-balancing strategy has to be applied when calling this particular dependency, you can provide it in the YAML file, and it is automatically applied. You can choose one of the following load balancing strategies:

  • STICKY: Once chosen, the instance is always called.

  • RANDOM: Picks an instance randomly.

  • ROUND_ROBIN: Iterates over instances over and over again.

5.3.4. Content-Type Template and Version

The Content-Type template and version are represented by the contentTypeTemplate and version YAML properties.

If you version your API in the Content-Type header, you do not want to add this header to each of your requests. Also, if you want to call a new version of the API, you do not want to roam around your code to bump up the API version. That is why you can provide a contentTypeTemplate with a special $version placeholder. That placeholder will be filled by the value of the version YAML property. Consider the following example of a contentTypeTemplate:

application/vnd.newsletter.$version+json

Further consider the following version:

v1

The combination of contentTypeTemplate and version results in the creation of a Content-Type header for each request, as follows:

application/vnd.newsletter.v1+json

5.3.5. Default Headers

Default headers are represented by the headers map in YAML.

Sometimes, each call to a dependency requires setting up of some default headers. To not do that in code, you can set them up in the YAML file, as shown in the following example headers section:

headers:
    Accept:
        - text/html
        - application/xhtml+xml
    Cache-Control:
        - no-cache

That headers section results in adding the Accept and Cache-Control headers with appropriate list of values in your HTTP request.

5.3.6. Required Dependencies

Required dependencies are represented by required property in YAML.

If one of your dependencies is required to be up when your application boots, you can set the required: true property in the YAML file.

If your application cannot localize the required dependency during boot time, it throws an exception, and the Spring Context fails to set up. In other words, your application cannot start if the required dependency is not registered in Zookeeper.

You can read more about Spring Cloud Zookeeper Presence Checker later in this document.

5.3.7. Stubs

You can provide a colon-separated path to the JAR containing stubs of the dependency, as shown in the following example:

stubs: org.springframework:myApp:stubs

where:

  • org.springframework is the groupId.

  • myApp is the artifactId.

  • stubs is the classifier. (Note that stubs is the default value.)

Because stubs is the default classifier, the preceding example is equal to the following example:

stubs: org.springframework:myApp

5.4. Configuring Spring Cloud Zookeeper Dependencies

You can set the following properties to enable or disable parts of Zookeeper Dependencies functionalities:

  • spring.cloud.zookeeper.dependencies: If you do not set this property, you cannot use Zookeeper Dependencies.

  • spring.cloud.zookeeper.dependency.loadbalancer.enabled (enabled by default): Turns on Zookeeper-specific custom load-balancing strategies, including ZookeeperServiceInstanceListSupplier and dependency-based load-balanced RestTemplate setup.

  • spring.cloud.zookeeper.dependency.headers.enabled (enabled by default): This property registers a FeignBlockingLoadBalancerClient that automatically appends appropriate headers and content types with their versions, as presented in the Dependency configuration. Without this setting, those two parameters do not work.

  • spring.cloud.zookeeper.dependency.resttemplate.enabled (enabled by default): When enabled, this property modifies the request headers of a @LoadBalanced-annotated RestTemplate such that it passes headers and content type with the version set in dependency configuration. Without this setting, those two parameters do not work.

6. Spring Cloud Zookeeper Dependency Watcher

The Dependency Watcher mechanism lets you register listeners to your dependencies. The functionality is, in fact, an implementation of the Observator pattern. When a dependency changes, its state (to either UP or DOWN), some custom logic can be applied.

6.1. Activating

Spring Cloud Zookeeper Dependencies functionality needs to be enabled for you to use the Dependency Watcher mechanism.

6.2. Registering a Listener

To register a listener, you must implement an interface called org.springframework.cloud.zookeeper.discovery.watcher.DependencyWatcherListener and register it as a bean. The interface gives you one method:

void stateChanged(String dependencyName, DependencyState newState);

If you want to register a listener for a particular dependency, the dependencyName would be the discriminator for your concrete implementation. newState provides you with information about whether your dependency has changed to CONNECTED or DISCONNECTED.

6.3. Using the Presence Checker

Bound with the Dependency Watcher is the functionality called Presence Checker. It lets you provide custom behavior when your application boots, to react according to the state of your dependencies.

The default implementation of the abstract org.springframework.cloud.zookeeper.discovery.watcher.presence.DependencyPresenceOnStartupVerifier class is the org.springframework.cloud.zookeeper.discovery.watcher.presence.DefaultDependencyPresenceOnStartupVerifier, which works in the following way.

  1. If the dependency is marked us required and is not in Zookeeper, when your application boots, it throws an exception and shuts down.

  2. If the dependency is not required, the org.springframework.cloud.zookeeper.discovery.watcher.presence.LogMissingDependencyChecker logs that the dependency is missing at the WARN level.

Because the DefaultDependencyPresenceOnStartupVerifier is registered only when there is no bean of type DependencyPresenceOnStartupVerifier, this functionality can be overridden.

7. Distributed Configuration with Zookeeper

Zookeeper provides a hierarchical namespace that lets clients store arbitrary data, such as configuration data. Spring Cloud Zookeeper Config is an alternative to the Config Server and Client. Configuration is loaded into the Spring Environment during the special “bootstrap” phase. Configuration is stored in the /config namespace by default. Multiple PropertySource instances are created, based on the application’s name and the active profiles, to mimic the Spring Cloud Config order of resolving properties. For example, an application with a name of testApp and with the dev profile has the following property sources created for it:

  • config/testApp,dev

  • config/testApp

  • config/application,dev

  • config/application

The most specific property source is at the top, with the least specific at the bottom. Properties in the config/application namespace apply to all applications that use zookeeper for configuration. Properties in the config/testApp namespace are available only to the instances of the service named testApp.

Configuration is currently read on startup of the application. Sending a HTTP POST request to /refresh causes the configuration to be reloaded. Watching the configuration namespace (which Zookeeper supports) is not currently implemented.

7.1. Activating

Including a dependency on org.springframework.cloud:spring-cloud-starter-zookeeper-config enables autoconfiguration that sets up Spring Cloud Zookeeper Config.

When working with version 3.4 of Zookeeper you need to change the way you include the dependency as described here.

7.2. Customizing

Zookeeper Config may be customized by setting the following properties:

bootstrap.yml
spring:
  cloud:
    zookeeper:
      config:
        enabled: true
        root: configuration
        defaultContext: apps
        profileSeparator: '::'
  • enabled: Setting this value to false disables Zookeeper Config.

  • root: Sets the base namespace for configuration values.

  • defaultContext: Sets the name used by all applications.

  • profileSeparator: Sets the value of the separator used to separate the profile name in property sources with profiles.

7.3. Access Control Lists (ACLs)

You can add authentication information for Zookeeper ACLs by calling the addAuthInfo method of a CuratorFramework bean. One way to accomplish this is to provide your own CuratorFramework bean, as shown in the following example:

@BoostrapConfiguration
public class CustomCuratorFrameworkConfig {

  @Bean
  public CuratorFramework curatorFramework() {
    CuratorFramework curator = new CuratorFramework();
    curator.addAuthInfo("digest", "user:password".getBytes());
    return curator;
  }

}

Consult the ZookeeperAutoConfiguration class to see how the CuratorFramework bean’s default configuration.

Alternatively, you can add your credentials from a class that depends on the existing CuratorFramework bean, as shown in the following example:

@BoostrapConfiguration
public class DefaultCuratorFrameworkConfig {

  public ZookeeperConfig(CuratorFramework curator) {
    curator.addAuthInfo("digest", "user:password".getBytes());
  }

}

The creation of this bean must occur during the boostrapping phase. You can register configuration classes to run during this phase by annotating them with @BootstrapConfiguration and including them in a comma-separated list that you set as the value of the org.springframework.cloud.bootstrap.BootstrapConfiguration property in the resources/META-INF/spring.factories file, as shown in the following example:

resources/META-INF/spring.factories
org.springframework.cloud.bootstrap.BootstrapConfiguration=\
my.project.CustomCuratorFrameworkConfig,\
my.project.DefaultCuratorFrameworkConfig