Skip to content

Latest commit

 

History

History
583 lines (442 loc) · 19.1 KB

README.md

File metadata and controls

583 lines (442 loc) · 19.1 KB

KumuluzEE Health

Build Status

KumuluzEE Health project provides consistent, unified way of performing microservice health checks and exposing health information.

KumuluzEE Health is a health check project for the KumuluzEE microservice framework. It provides easy, consistent and unified way of performing health checking on microservices and exposing health information to be used by monitoring and container orchestration environments such as Kubernetes. KumuluzEE Health is fully compliant with Kubernetes and has been extensively tested to work in Kubernetes.

KumuluzEE Health is compliant with the MicroProfile Service Health Checks specification 2.2.

KumuluzEE Health exposes /health/live and /health/ready endpoints (prefix customizable), which return the health check status of the microservice.

Usage

You can enable the KumuluzEE Health module by adding the following dependencies:

<dependency>
    <groupId>com.kumuluz.ee.health</groupId>
    <artifactId>kumuluzee-health</artifactId>
    <version>${kumuluzee-health.version}</version>
</dependency>

CDI and JAX-RS dependencies are a prerequisite. Please refer to KumuluzEE readme for more information.

Health checks

To check health of a microservice, you can use the provided health checks or you can define your own health checks.

Liveness and readiness

KumuluzEE Health differentiates between two health check types - liveness and readiness health check. In short - if a liveness check fails it means that the service is stuck and should be restarted. If a readiness check fails it means that the service is temporary unavailable and should not receive requests until all readiness checks succeed.

For more information on liveness and readiness see the following Kubernetes articles:

Built-in health checks

The following health checks are available out-of-the-box:

  • DataSourceHealthCheck for checking the availability of the data source
  • DiskSpaceHealthCheck for checking available disk space against a threshold
  • ElasticSearchHealthCheck for checking the availability of Elasticsearch cluster
  • EtcdHealthCheck for checking the availability of etcd instance
  • HttpHealthCheck for checking the availability of HTTP resource
  • MongoHealthCheck for checking the availability of Mongo database
  • RabbitHealthCheck for checking the availability of RabbitMQ virtual host
  • RedisHealthCheck for checking the availability of Redis store
  • KafkaHealthCheck for checking the availability of Kafka cluster

More detailed descriptions of each health check are provided below. Additional built-in health check will be provided (contributions are welcome).

Implementing custom health checks

There are two ways how we can implement a custom health check.

  • We can use the @Liveness and @Readiness annotation to define health check classes.
  • We can implement health check classes and register them manually.

@Liveness and @Readiness annotation

To implement health checks using @Liveness or @Readiness annotation, we have to implement a CDI bean class which implements the HealthCheck interface. Such health checks are automatically discovered and registered to the HealthRegistry.

Shown below is an example of a CDI bean health check using @Readiness annotation:

import org.eclipse.microprofile.health.Readiness;
import org.eclipse.microprofile.health.HealthCheck;
import org.eclipse.microprofile.health.HealthCheckResponse;

import javax.enterprise.context.ApplicationScoped;

@Readiness
@ApplicationScoped
public class SuccessfulHealthCheckBean implements HealthCheck {

    public HealthCheckResponse call() {
        return HealthCheckResponse.up(SuccessfulHealthCheckBean.class.getSimpleName());
    }

}

A health check can also be annotated with both @Liveness and @Readiness at the same time.

Health check implemented as class

To implement a health check with a custom class, the class has to implement the HealthCheck interface. Such class has to be manually registered with the HealthRegistry.

Shown below is a custom health check implementation. It checks if the KumuluzEE GitHub page is accessible.

import org.eclipse.microprofile.health.HealthCheck;
import org.eclipse.microprofile.health.HealthCheckResponse;

import java.net.HttpURLConnection;
import java.net.URL;
import java.util.logging.Logger;

public class GithubHealthCheck implements HealthCheck {

    private static final String url = "https://github.com/kumuluz/kumuluzee";

    private static final Logger LOG = Logger.getLogger(GithubHealthCheck.class.getSimpleName());

    @Override
    public HealthCheckResponse call() {
        try {
            HttpURLConnection connection = (HttpURLConnection) new URL(url).openConnection();
            connection.setRequestMethod("HEAD");

            if (connection.getResponseCode() == 200) {
                return HealthCheckResponse.up(GithubHealthCheck.class.getSimpleName());
            }
        } catch (Exception exception) {
            LOG.severe(exception.getMessage());
        }
        return HealthCheckResponse.down(GithubHealthCheck.class.getSimpleName());
    }
}

Registering custom health checks

To register a custom health check class we have to use the HealthRegistry instance. We provide the health check unique name, an instance of the health check class ant the health check type (liveness/readiness).

HealthRegistry.getInstance().register(GithubHealthCheck.class.getSimpleName(), new GithubHealthCheck(), HealthCheckType.LIVENESS);

Unregistering custom health checks

To unregister custom health checks we can use the HealthRegistry instance and provide the health check unique name.

HealthRegistry.getInstance().unregister(GithubHealthCheck.class.getSimpleName());

Retrieving health check results

To invoke the health check and retrieve the result we can use the HealthRegistry instance. The results will be returned in a list of health check responses. The desired health check type must also be specified.

List<HealthCheckResponse> results = HealthRegistry.getInstance().getResults(HealthCheckType.BOTH);

/health/* endpoint output

The /health/live and /health/ready endpoints return:

  • 200 with payload, when health checks are defined with positive status or are not defined
  • 503 with payload, when health checks are defined, but at least one status is negative
  • 500 without payload, when an exception occurred in the procedure of health checking

The health check is available on http://IP:PORT/health/live and http://IP:PORT/health/ready by default, payload example is provided below:

{
  "status" : "UP",
  "checks" : [ {
    "name" : "DataSourceHealthCheck",
    "status" : "UP"
  }, {
    "name" : "DiskSpaceHealthCheck",
    "status" : "UP"
  }, {
    "name" : "ElasticSearchHealthCheck",
    "status" : "UP"
  }, {
    "name" : "EtcdHealthCheck",
    "status" : "UP",
    "data": {
      "http://localhost:2379": "UP" 
    }
  }, {
    "name" : "HttpHealthCheck",
    "status" : "UP",
    "data": {
      "https://github.com/kumuluz/kumuluzee-health": "UP"
    }
  }, {
    "name" : "MongoHealthCheck",
    "status" : "UP"
  }, {
    "name" : "RabbitHealthCheck",
    "status" : "UP"
  }, {
    "name" : "RedisHealthCheck",
    "status" : "UP"
  } ]
}

The URL also accepts a query parameter pretty=false (http://IP:PORT/health/ready?pretty=false) which results in a single line response, payload example is provided below:

{"status":"UP","checks":[{"name":"DataSourceHealthCheck","status":"UP"},{"name":"DiskSpaceHealthCheck","status":"UP"},{"name":"ElasticSearchHealthCheck","status":"UP"},{"name":"EtcdHealthCheck","status":"UP","data":{"http://localhost:2379": "UP"}},{"name":"HttpHealthCheck","status":"UP","data":{"https://github.com/kumuluz/kumuluzee-health":"UP"}},{"name":"MongoHealthCheck","status":"UP"},{"name":"RabbitHealthCheck","status":"UP"},{"name":"RedisHealthCheck","status":"UP"}]}

Configuring health check endpoint prefix

Health check is provided via URL, the health servlet is registered automatically on path /health/*. To configure the health check endpoint prefix, you can specify the following configuration keys:

  • kumuluzee.health.servlet.mapping: Health servlet path. Default value is /health/*.
  • kumuluzee.health.servlet.enabled: Is JSON output enabled. Default value is true. If false only the status codes will be provided.

The JSON output will also be enabled if the DEBUG mode is enabled, by setting kumuluz.debug to true.

Example of the configuration:

kumuluzee:
  health:
    servlet:
      mapping: /my-health
      enabled: true

Enabling health check logging

Periodic logging of health check results is also available. To configure the health check results logging, you can specify the following configuration keys:

  • kumuluzee.health.logs.enabled: Is logging enabled. Default value is true.
  • kumuluzee.health.logs.type: Type of health checks to be logged. Allowed values: both, readiness, liveness. Default value is both.
  • kumuluzee.health.logs.level: The logging level. Default value is FINE.
  • kumuluzee.health.logs.period-s: The logging period in seconds. Default value is 60.

Example of the configuration:

kumuluzee:
  health:
    logs:
      enabled: true
      type: readiness
      level: FINE
      period-s: 60

Disabling health extension

Health extension can be disabled by setting the configuration property kumuluzee.health.enabled to false. This will disable the /health endpoint and disable health check logging.

Configuring built-in health checks

To configure built-in health checks, we can use the configuration parameters listed below for each built-in health check. Every built-in health check supports a configuration key type which specifies under which type the health check should be registered. Allowed values are liveness, readiness and both. Default value is readiness. For example see the below configuration of DataSourceHealthCheck.

DataSourceHealthCheck

To enable data source availability health check, we need to provide in the health check sections. Jndi-name, connection-url, username and password need to be provided as part of the health check configuration. Note that multiple data source health checks are supported.

Example configuration:

kumuluzee:
  datasources:
    - jndi-name: jdbc/CustomersDS
      connection-url: jdbc:postgresql://localhost:5432/customers
      username: postgres
      password: postgres
      pool:
      	max-size: 20
    - jndi-name: jdbc/OrdersDS
      connection-url: jdbc:postgresql://localhost:5433/orders
      username: postgres
      password: postgres
      pool:
      	max-size: 20
  health:
    checks:
      data-source-health-checks
        - jndi-name: jdbc/CustomersDS
	  type: liveness
	- jndi-name: jdbc/OrdersDS
	  type: liveness

Another example of the configuration:

kumuluzee:
  health:
    checks:
      data-source-health-check:
        connection-url: jdbc:db2://localhost:5021/customers
        username: db2
        password: db2

Another example of the configuration:

kumuluzee:
  health:
    checks:
      data-source-health-check:
        connection-url: jdbc:mysql://localhost:3306/customers?user=mysql&password=mysql

To enable data source availability health check, we also need to provide a database driver library in pom.xml.

Example configuration:

<dependency>
	<groupId>org.postgresql</groupId>
	<artifactId>postgresql</artifactId>
	<version>42.0.0</version>
</dependency>

DiskSpaceHealthCheck

To enable disk space health check, we need to provide the health check config parameters, listed below. The default disk space threshold is 100MB, but can be overwritten by providing your own threshold.

Example of the configuration:

kumuluzee:
  health:
    checks:
      disk-space-health-check:
        threshold: 100000000

ElasticSearchHealthCheck

To enable Elasticsearch cluster health check, we need to specify the connection-url with cluster health check endpoint as part of the health check configuration. The cluster health check endpoint is typically available on http://HOST:IP/_cluster/health. The response should resemble:

{
  "cluster_name" : "testcluster",
  "status" : "yellow",
  "timed_out" : false,
  "number_of_nodes" : 1,
  "number_of_data_nodes" : 1,
  "active_primary_shards" : 5,
  "active_shards" : 5,
  "relocating_shards" : 0,
  "initializing_shards" : 0,
  "unassigned_shards" : 5,
  "delayed_unassigned_shards": 0,
  "number_of_pending_tasks" : 0,
  "number_of_in_flight_fetch": 0,
  "task_max_waiting_in_queue_millis": 0,
  "active_shards_percent_as_number": 50.0
}

ElasticSearchHealthCheck checks if the status of HTTP response is 200 and if status field is either green or yellow. The default connection-url is http://localhost:9200/_cluster/health.

Example of the configuration:

kumuluzee:
  health:
    checks:
      elastic-search-health-check:
        connection-url: http://localhost:9200/_cluster/health?pretty

EtcdHealthCheck

To enable etcd health check, we need to specify the connection-url or multiple connection-url as part of the health check configuration.

Example configuration:

kumuluzee:
  health:
    checks:
      etcd-health-check:
        connection-url: http://localhost:2379/health

Another example of the configuration:

kumuluzee:
  health:
    checks:
      etcd-health-check:
        - connection-url: http://localhost:2379/health
        - connection-url: http://192.168.99.100:2379/health

HttpHealthCheck

We can provide single or multiple urls for HTTP availability health check. To enable HTTP availability health check, we need to specify the connection-url or multiple connection-url as part of the health check configuration. During the http health check HEAD requests are made to all the connection-url and status code is verified if its >=200 and <300.

Example configuration:

kumuluzee:
  health:
    checks:
      http-health-check:
        connection-url: https://github.com/kumuluz/kumuluzee-health

Another example of the configuration:

kumuluzee:
  health:
    checks:
      http-health-check:
        - connection-url: https://github.com/kumuluz/kumuluzee-health
        - connection-url: http://www.reddit.com

MongoHealthCheck

To enable the Mongo database health check, we need to provide the connection-url config parameter with user, password, database name and other options need to be provided as part of the health check configuration as described in the mongo-java-driver-documentation. The default connection-url is mongodb://localhost:27017/local?serverSelectionTimeoutMS=2000.

Example of the configuration:

kumuluzee:
  health:
    checks:
      mongo-health-check:
        connection-url: mongodb://user:password@localhost:27017/customers?serverSelectionTimeoutMS=2000

To enable the Mongo database health check, we also need to provide mongo-java-driver library in pom.xml.

Example configuration:

<dependency>
	<groupId>org.mongodb</groupId>
	<artifactId>mongo-java-driver</artifactId>
	<version>3.9.1</version>
</dependency>

RabbitHealthCheck

To enable RabbitMQ health check, we need to specify the connection-url with port, username, password and virtual host as part of the health check configuration. The default connection-url is amqp://guest:guest@localhost:5672?connection_timeout=2000.

Example of the configuration:

kumuluzee:
  health:
    checks:
      rabbit-health-check:
        connection-url: amqp://guest:guest@localhost:5672/virtualHost?connection_timeout=2000

To enable RabbitMQ health check, we also need to provide amqp-client library in pom.xml.

Example configuration:

<dependency>
	<groupId>com.rabbitmq</groupId>
	<artifactId>amqp-client</artifactId>
	<version>5.6.0</version>
</dependency>

RedisHealthCheck

To enable Redis store health check, we need to specify the connection-url with port, secret and database number as part of the health check configuration. The default connection-url is redis://localhost:6379/0.

Example of the configuration:

kumuluzee:
  health:
    checks:
      redis-health-check:
        connection-url: redis://:secret@localhost:6379/0

To enable Redis store health check, we also need to provide jedis library in pom.xml.

Example configuration:

<dependency>
	<groupId>redis.clients</groupId>
	<artifactId>jedis</artifactId>
	<version>3.0.1</version>
</dependency>

KafkaHealthCheck

To enable Kafka cluster health check provide the bootstrap-servers urls as a comma separated list. The default value is localhost:9092.

You can supply the minimum-available-nodes value (default is 1). The value represents the minimum number of available nodes in the cluster in order for the health check to be considered successful.

Example of the configuration:

kumuluzee:
  health:
    checks:
      kafka-health-check:
        bootstrap-servers: localhost:9095,localhost:9096,localhost:9097
        minimum-available-nodes: 2
        request-timeout-ms: 1000

Additionally, all properties defined in the subtree kumuluzee.health.checks.kafka-health-check are forwarded to Kafka client (see CONFGIGURATION). Note that dots (.) are replaced with minuses (-), e.g. request.timeout.ms becomes request-timeout-ms.

The following dependency needs to be provided in order for the health check to function correctly (when using kumuluzee-streaming library the dependency is already provided):

<dependency>
    <groupId>org.apache.kafka</groupId>
    <artifactId>kafka-clients</artifactId>
    <version>2.6.0</version>
</dependency>

Changelog

Recent changes can be viewed on Github on the Releases Page

NOTE FOR 1.x USERS: The /health endpoint and the @Health annotation are deprecated as of 2.0.0 release. They still work as before but should not be used for any new services. Also note that property names of the JSON structure were changed in the 2.0.0 release.

Contribute

See the contributing docs

When submitting an issue, please follow the guidelines.

When submitting a bugfix, write a test that exposes the bug and fails before applying your fix. Submit the test alongside the fix.

When submitting a new feature, add tests that cover the feature.

License

MIT