Spring Batch Lightmin is a client/server batch and scheduling platform for the Spring Boot stack. The batch stack based on Spring Batch.
All samples for Spring Batch Lightmin can be found at Sample Applications
|
Important
|
Spring 5.1.x set the the property spring.main.allow-bean-definition-overriding to false as a default. In order to use the current version of Spring Batch Lightmin, this property has to be set to true. This issue can hopefully be solved in the upcoming releases. |
The following section describes a quick start guide to setup the client server architecture. If your want to get an overview of the configuration options, please read the documentation. The getting started guide for the Embedded Mode can be found in the section Embedded Lightmin.
Add the Spring Batch Lightmin BOM to the dependency management section of your project.
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-bom</artifactId>
<version>${spring-batch-lightmin.version}</version>
<type>pom</type>
<scope>import</scope>
</dependency>The server is the component which is used to configure clients and schedule jobs.
|
Important
|
The server can be operated in standalone and cluster mode. The standalone mode does not support any synchronisation between multiple cluster instances. Also the server cluster scheduler are working totally independent from each other in standalone mode. |
The server itself is also a Spring Boot application.
The following dependency has to added to your project.
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-server-standalone</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>In order to enable the server, the annotation @EnableLightminServer has to added to one of the @Configuration classes.
@EnableLightminServer
@SpringBootApplication
public class LightminServer {
}If you added the cluster dependencies the following actions have to be done:
In order to enable the server, the annotation @EnableLightminServer has to added to one of the @Configuration classes.
@EnableLightminServer
@EnableServerClusterInfinispan
@SpringBootApplication
public class LightminServer {
}The Infinispan cluster configuration will provide the following clustered features:
-
Synchronized poller for server scheduler and clean up jobs
-
LightminServerLockManager implenentation for Infinispan
-
Distributed Repositories for
-
JobExecutionEventRepository
-
JournalRepository
-
LightminApplicationRepository
-
If the following properties are set to true, additional distributed repositories will be provided:
spring:
batch:
lightmin:
server:
cluster:
infinispan:
repository:
init-scheduler-configuration-repository: true # default false (1)
init-scheduler-execution-repository: true # default false (2)-
if true, a distributed SchedulerConfigurationRepository will be provided
-
if true, a distributed SchedulerExecutionRepository will be provided
|
Important
|
It is highly recommended to use persistent repositories for SchedulerConfigurationRepository and SchedulerExecutionRepository like the JDBC repositories. The distributed repositories will contain the stored values as long as one cluster member is up and running. |
In order to use the infinispan cluster, a GlobalConfiguration has to be provided. The documentation of Infinispan and jGroups will give you insides how to configure it for your need.
@Configuration
public class MyConfiguration {
@Bean
public GlobalConfiguration globalConfiguration() {
return new GlobalConfigurationBuilder()
// Configure the global configuration
.build();
}
}Since lightmin version 2.1.1 the server provides a scheduler to trigger jobs at the clients. In order to provide this feature, the scheduler configurations and executions have to be stored.
The SchedulerConfigurationRepository and SchedulerExecutionRepository are responsible for the doing the job. Per dafault, the standalone server will configure inmemory repositories, which are cleared after each restart.
The behavior of the Server Scheduler can be changed via application properties.
spring:
batch:
lightmin:
server:
scheduler:
enabled: true # default true (1)
create-new-executions-on-failure: false # default false (2)
create-new-executions-on-lost: false # default false (3)
fail-on-instance-execution-count: true # default true (4)
poller-period: 1000 # default 1000 (5)
poller-period-retry: 1000 # default 1000 (6)
repository:
delete-poller-period: PT10m # default 10 minutes (7)
delete-failed: true # default true (8)
delete-finished: true # default true (9)
delete-lost: true # default true (10)
keep-failed: PT24h # default 24 hours (11)
keep-finished: PT24h # default 24 hours (12)
keep-lost: PT24h # default 24 hours (13)-
If set to false, no schedulers will be executed
-
If set to true, a new executions based on the configuration will be created on failure.
-
If set to true, a new executions based on the configuration will be created on lost after configured retries failed.
-
If true, the execution will fail if not enough instances are available.
-
Time in millis to fetch new executions to be executed by the scheduler
-
Time in millis to fetch executions to be retried by the scheduler
-
Duration of the clean up poller for the repositories
-
If set to true, the failed executions will be deleted automatically.
-
If set to true, the finished executions will be deleted automatically.
-
If set to true, the lost executions will be deleted automatically.
-
Duration how long failed executions should be kept before getting deleted automatically.
-
Duration how long finished executions should be kept before getting deleted automatically.
-
Duration how long lost executions should be kept before getting deleted automatically.
|
Important
|
The properties create-new-executions-on-failure: false create-new-executions-on-lost: false fail-on-instance-execution-count: true are global properties. There is the possibility that future releases will support this configuration option per execution. |
The database schema for the repositories are available in the dependency at org.tuxdevelop.spring.batch.lightmin.server.scheduler.repository
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-server-scheduler-repository-jdbc</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>Depending on the clients, the server is also able to find the clients via service discovery.
To enable the feature, a DiscoveryClient implementation has to added and configured(e.g. Consul, Eureka, etc…) and the following property has to set to true
spring::
batch:
lightmin:
server:
discovery-enabled: trueThe server frontend uses redirects and the HOST header is taken to create the redirect urls. This is the default behaviour of Spring MVC. If the server frontend is running behind a proxy, the implementation of the server takes care, that the X-FORWARD-PREFIX header is used as well.
If the proxy cannot pass the HOST header or you do not want to change the defaults, e.g. Zuul, the following property can force the server to use the X-FORWARDED-HOST header to build the redirect urls.
spring:
batch:
lightmin:
server:
use-x-forwarded-headers: trueThe client applications are responsible to provide Spring Batch Job definitions. The Spring Batch Lightmin client framework provides all the configurations to set up Spring Batch and the communication with the server.
The type of the client decides how the registration to the server should be done. Currently to types are supported, classic and via service discovery.
The classic client has to know where the servers are located and will send a registration request after the start up.
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-client-classic</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The dependency above will provide everything which is required for the classic client. The annotation @EnableLightminClientClassic has to added to one of the configuration classes.
@SpringBootApplication
@EnableLightminClientClassic
public class ClientApplication {
public static void main(final String[] args) {
SpringApplication.run(ClientApplication.class, args);
}
}The following configuration properties have to be present
spring:
application:
name: my-client-application (1)
batch:
lightmin:
client:
classic:
server:
url: http://myserver1.domain:8180, http://myserver2.domain:8180 (2)-
The spring.application.name is used to identify a client and handle a cluster of the instances.
-
The url property is a list of server to which the registration request should be send.
Spring Batch Lightmin provides two implementations for the discovery client type. Both implementations add a tags to the underlying service discovery technology, so the server could identify lightmin clients.
The following dependency has to added for the consul client
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-client-discovery-consul</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The annotation @EnableLightminClientConsul enables the fully integration with Consul.
@SpringBootApplication
@EnableLightminClientConsul
public class ClientApplication {
public static void main(final String[] args) {
SpringApplication.run(ClientApplication.class, args);
}
}The configuration options for Consul can be found in the Spring Cloud Consul documentation.
The client is sending per default events to the server, in order to find the server via service discovery, the following property has to be set.
spring:
batch:
lightmin:
client:
discovery:
server-discovery-name: lightmin-server (1)-
The service discovery name of the server.
The following dependency has to added for the eureka client
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-client-discovery-eureka</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The annotation @EnableLightminClientEureka enables the fully integration with Eureka.
@SpringBootApplication
@EnableLightminClientEureka
public class ClientApplication {
public static void main(final String[] args) {
SpringApplication.run(ClientApplication.class, args);
}
}The configuration options for Eureka can be found in the Spring Cloud Netflix documentation.
The client is sending per default events to the server, in order to find the server via service discovery, the following property has to be set.
spring:
batch:
lightmin:
client:
discovery:
server-discovery-name: lightmin-server (1)-
The service discovery name of the server.
The Configuration Repository is the component which stores the scheduler and listener configurations of the lightmin clients. This configurations are loaded at start time and can be managed with the server frontend or API calls.
Spring Batch Lightmin provides three implementation of the repository
-
map - In memory repository, all changes will be gone after a restart.
-
jdbc - The client fetches and stores the configurations in a database.
-
remote - The client fetches and stores the configurations via API calls to a repository server.
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-repository-map</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The annotation @EnableLightminMapConfigurationRepository enables the in memory repository.
@SpringBootApplication
@EnableLightminClientConsul
@EnableLightminMapConfigurationRepository
public class ClientApplication {
public static void main(final String[] args) {
SpringApplication.run(ClientApplication.class, args);
}
}<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-repository-jdbc</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The annotation @EnableLightminJdbcConfigurationRepository enables the jdbc repository.
@SpringBootApplication
@EnableLightminClientConsul
@EnableLightminJdbcConfigurationRepository
public class ClientApplication {
public static void main(final String[] args) {
SpringApplication.run(ClientApplication.class, args);
}
}The jdbc repository requires a configured datasource bean with the name dataSource.
If the project configuration requires a specific datasource for the lightmin repository, the bean name can set via configuration property. More configuration options can be found in the documentation.
spring:
batch:
lightmin:
repository:
jdbc:
data-source-name: myDataSource (1)-
Overriding the default datasource name.
The database schema ddl and drop scripts for various databases are located in the dependency above under the path:
org/tuxdevelop/spring/batch/lightmin/repository<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-repository-remote</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The annotation @EnableLightminRemoteConfigurationRepository enables the remote repository.
@SpringBootApplication
@EnableLightminClientConsul
@EnableLightminRemoteConfigurationRepository
public class ClientApplication {
public static void main(final String[] args) {
SpringApplication.run(ClientApplication.class, args);
}
}The remote repository can be located via url and service discovery.
For the url way, the following properties has to be set:
spring:
batch:
lightmin:
repository:
remote:
server-url: http://my-server.domain:8280 (1)-
The url to the remote repository server
For the service discovery approach, the following properties has to set and a DiscoveryClient bean has to be present.
spring:
batch:
lightmin:
repository:
remote:
discover-remote-repository: true (1)
server-discovery-name: remoteRepositorySever (2)-
Enables the discovery feature
-
The discovery name of the remote repository server
Further configuration options can be found in the documentation.
The client configurations are enabling the Spring Batch stack as well. Spring Batch itself has to have a configured JobRepository. This JobRepository can be in memory via map or jdbc.
The configuration of the JobRepository can be done via properties, so Spring Batch Lightmin knows what to configure.
For the map repository, the following configuration is enough:
spring:
batch:
lightmin:
batch:
repository-type: mapIf a client runs inside a container like Docker, the dns name of the host systems has to be transferred to server on registration time. For this use case, a property is available.
spring:
batch:
lightmin:
client:
hostname: FQDN of the hostThe Remote Repository Server is a Spring Boot application which provides a REST API for clients.
The server itself needs a job configuration repository itself. The server supports map and jdbc.
For the Jdbc repository, the following dependencies have to be added.
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-repository-server</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-repository-jdbc</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>The configuration of the used repository can be found in the client section.
The annotation @EnableLightminRepositoryServer enables the server and the corresponding annotation the job configuration repository.
@SpringBootApplication
@EnableLightminRepositoryServer
@EnableLightminJdbcConfigurationRepository
public class RepositoryServerApplication {
public static void main(final String[] args) {
SpringApplication.run(RepositoryServerApplication.class, args);
}
}If the client server architecture does not fit the requirements, Spring Batch Lightmin also provides am embedded mode, which ships the client and server in one package.
<dependency>
<groupId>org.tuxdevelop</groupId>
<artifactId>spring-batch-lightmin-embedded</artifactId>
<version>${spring-batch-lightmin.version}</version>
</dependency>In this case, a specific client does not have to be configured. The Job Configuration Repository and Spring Batch have to be configured like for a regular client.
@SpringBootApplication
@EnableLightminEmbedded
@EnableLightminMapConfigurationRepository
public class EmbeddedLightminApplication {
public static void main(final String[] args) {
SpringApplication.run(EmbeddedLightminApplication.class, args);
}
}spring:
batch:
lightmin:
batch:
repository-type: mapLightmin Metrics introduces some custom metrics for the Lightmin-Platform. It uses Micrometer to collect metrics. The metrics are available for client and server applications.
In order to collect metrics of a Lightmin client application, the following property has to be to true.
spring:
batch:
lightmin:
client:
metrics-enabled: true # default true|
Important
|
If you want to use runtime information of a client on a server, the following property has to be set to true. The information send will be transformed to metrics on the server, if the sever metrics feature is enabled. |
spring:
batch:
lightmin:
client:
publish-job-events: true # default trueIn order to collect metrics of a Lightmin server , the following property has to be set to true.
spring:
batch:
lightmin:
server:
metrics-enabled: true # default trueIn order to expose the collected metrics, a Micrometer registry dependencies has to be added to the client and the server.
For prometheus, the following dependencies has to be added
<dependency>
<groupId>io.micrometer</groupId>
<artifactId>micrometer-registry-prometheus</artifactId>
<version>${micrometer.version}</version>
</dependency>|
Important
|
The lightmin dependencies do not provide any registries out of the box |
Metrics will be published over prometheus actuator and can be modified by adding MeterFilter into the metrics configuration. Refer to Micrometer Documentation for more information.
@Configuration
public class Config{
final static String name = "EXAMPLE";
@Bean
public MeterRegistryCustomizer<MeterRegistry> metricsCommonTags() {
return registry -> {
if (name != null) {
registry.config().meterFilter(
new MeterFilter() {
@Override
public DistributionStatisticConfig configure(Meter.Id id, DistributionStatisticConfig config) {
if (id.getName().startsWith("lightmin")) {
// Example Filter - Activating histogram on all lightmin metrics
return DistributionStatisticConfig.builder().percentilesHistogram(true)
.percentiles(0.95)
.build()
.merge(config);
}
return config;
}
})
.commonTags("example_tag", name);
}
};
}
}Lightmin Metrics introduces prebuild Grafana Dashboards for server and client. The json files are available in the resources folder of the lightmin-metrics project. In order to use the plugin, please install the following Plugins:
-
Statusmap by Flant JSC
The Spring Batch Lightmin Dashboard provides information about all Lightmin clients which are configured as target in Prometheus. The following variables are available:
-
Service - taken from the tag APPLICATION_NAME (Lightmin Clients should be listed here)
-
Job Name - Name of all jobs, or all jobs of the selected Services
-
Job Count Interval - Time period for the overall SUCCESSFUL and FAILED overview
If a specific Job Name is selected, the details about the job and step execution will be displayed.
The start page of the SpringBatchLightmin shows all register applications. The status icon shows the current health status of the application.
The application dashboard is the entry point to the monitoring and administration of a client application instance. The overview shows the important endpoints, all known Spring Batch Jobs and configured external links of the client application.
The batch jobs overview shows all registered batch jobs of the application instance and the execution count of them.
The view shows an overview of all executions for the selected job. To get details of the job execution, click on the desired id.
The job execution view shows you a detailed overview about the job and step executions of the selected job execution.
Job Scheduler Configurations are cron or time based scheduler.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent