Quartz

Configuring Quartz with JDBCJobStore in Spring

I am starting a little series about Quartz scheduler internals, tips and tricks, this is a chapter 0 - how to configure persistent job store. In Quartz you essentially have a choice between storing jobs and triggers in memory and in a relation database ( Terracotta is a recent addition to the mix). I would say in 90% of the cases when you use RAMJobStore with Quartz you don't really need Quartz at all. Obviously this storage backend is transient and all your pending jobs and triggers are lost between restarts. If you are fine with that, much simpler and more lightweight solutions are available, including ScheduledExecutorService built into JDK and @Scheduled(cron="*/5 * * * * MON-FRI") in Spring. Can you justify using extra 0,5MiB JAR in this scenario?

This changes dramatically when you need clustering, fail-over, load-balancing and few other buzz-words. There are several use-cases for that:

• single server cannot handle required number of concurrent, long running jobs and the executions need to be split into several machines - but each task must be executed exactly ones
• we cannot afford to run jobs too late - if one server is down, another should run the job on time
• ...or less strictly - the job needs to run eventually - even if the one and only server was down for maintenance, delayed jobs need to be run as soon as possible after restart

In all cases above we need some sort of non-transient global storage to keep track which jobs were executed, so that they are run exactly ones by one machine. Relational database as a shared memory works good in this scenario.

So if you think you need to schedule jobs and have some of the requirements above, keep reading. I will show you how to configure Quartz with Spring and fully integrate them. First of all we need a DataSource:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

import org.apache.commons.dbcp.BasicDataSource import com.googlecode.flyway.core.Flyway import org.jdbcdslog.DataSourceProxy import org.springframework.jdbc.datasource.{DataSourceTransactionManager, LazyConnectionDataSourceProxy} import org.h2.Driver    @Configuration @EnableTransactionManagement class Persistence {        @Bean     def transactionManager() = new DataSourceTransactionManager(dataSource())        @Bean     @Primary     def dataSource() = {         val proxy = new DataSourceProxy()         proxy.setTargetDSDirect(dbcpDataSource())         new LazyConnectionDataSourceProxy(proxy)     }        @Bean(destroyMethod = "close")     def dbcpDataSource() = {         val dataSource = new BasicDataSource         dataSource.setDriverClassName(classOf[Driver].getName)         dataSource.setUrl("jdbc:h2:mem:quartz-demo;DB_CLOSE_DELAY=-1;DB_CLOSE_ON_EXIT=FALSE;MVCC=TRUE")         dataSource.setUsername("sa")         dataSource.setPassword("")         dataSource.setMaxActive(20)         dataSource.setMaxIdle(20)         dataSource.setMaxWait(10000)         dataSource.setInitialSize(5)         dataSource.setValidationQuery("SELECT 1")         dataSource     }    }

As you might have guessed, Quartz needs some database tables to work with. It does not create them automatically, but SQL scripts for several databases are provided, including H2 which as you can see I am using. I think Flyway is the easiest way to run database scripts on startup:

?

1 2 3 4 5 6

@Bean(initMethod = "migrate") def flyway() = {  val fly = new Flyway()  fly.setDataSource(dataSource())  fly }

BTW in case you haven't noticed: no, there is no XML in our sample application and yes, we are using Spring.

Let's move on to Quartz:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

@Configuration class Scheduling {        @Resource     val persistence: Persistence = null        @Bean     @DependsOn(Array("flyway"))     def schedulerFactory() = {         val schedulerFactoryBean = new SchedulerFactoryBean()         schedulerFactoryBean.setDataSource(persistence.dataSource())         schedulerFactoryBean.setTransactionManager(persistence.transactionManager())         schedulerFactoryBean.setConfigLocation(new ClassPathResource("quartz.properties"))         schedulerFactoryBean.setJobFactory(jobFactory())         schedulerFactoryBean.setApplicationContextSchedulerContextKey("applicationContext")         schedulerFactoryBean.setSchedulerContextAsMap(Map().asJava)         schedulerFactoryBean.setWaitForJobsToCompleteOnShutdown(true)         schedulerFactoryBean     }        @Bean     def jobFactory() = new SpringBeanJobFactory    }

It is nice to know you can inject instance of @Configuration annotated classes into another such class for convenience. Except that - nothing fancy. Note that we need @DependsOn(Array("flyway")) on Quartz scheduler factory - otherwise the scheduler might start before Flyway fired the migration script with Quartz database tables causing unpleasant errors on startup. The essential bits are SpringBeanJobFactory and schedulerContextAsMap. The special factory makes Spring responsible for creating Job instances. Unfortunately this factory is quite limited which we will see shortly in the following example. First we need a Spring bean and a Quartz job:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

@Service class Printer extends Logging {        def print(msg: String) {         logger.info(msg)     }    }    class PrintMessageJob extends Job with Logging {        @BeanProperty     var printer: Printer = _        @BeanProperty     var msg = ""        def execute(context: JobExecutionContext) {         printer print msg     } }

First unexpected input is @BeanProperty instead of @Autowired or @Resource. Turns out that Job is not really a Spring bean, even though Spring creates an instance of it. Instead Spring discovers required dependencies using available setters. So where does the msg string come from? Keep going:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

val trigger = newTrigger().         withIdentity("Every-few-seconds", "Demo").         withSchedule(             simpleSchedule().                     withIntervalInSeconds(4).                     repeatForever()         ).         build()    val job = newJob(classOf[PrintMessageJob]).         withIdentity("Print-message", "Demo").         usingJobData("msg", "Hello, world!").         build()    scheduler.scheduleJob(job, trigger)

Quartz 2.0 ships with a nice internal DSL for creating jobs and triggers in a readable manner. As you can see I am passing an extra "Hello, world!" parameter to the job. This parameter is stored in so called JobData in the database per job or per trigger. It will be provided to the job when it is executed. This way you can parametrize your jobs. However when executed our job throws NullPointerException... Apparently printer reference was not set and silently ignored. Turns out Spring won't simply look through all the beans available in the ApplicationContext. The SpringBeanJobFactory only looks into Jobs' and Triggers' JobData and into so called scheduler context (already mentioned). If you want to inject any Spring bean into Job you must explicitly place a reference to that bean in schedulerContext:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

@Configuration class Scheduling {        @Resource     val printer: Printer = null        @Bean     def schedulerFactory() = {         val schedulerFactoryBean = new SchedulerFactoryBean()         //...         schedulerFactoryBean.setSchedulerContextAsMap(schedulerContextMap().asJava)         //...         schedulerFactoryBean     }        def schedulerContextMap() =         Map(             "printer" -> printer         )    }

Unfortunately each and every Spring bean you want to inject to job has to be explicitly referenced in schedulerContextMap. Even worse, if you forget about it, Quartz will silently log NPE at runtime. In the future we will write more robust job factory. But for starters we have a working Spring + Quartz application ready for further experiments, sources as always available under my GitHub account.

You might ask yourself way can't we simply use MethodInvokingJobDetailFactoryBean? Well, first of all because it does not work with persistent job stores. Secondly - because it is unable to pass JobData to Job - so we can no longer distinguish between different job runs. For instance our job printing message would have to always print the same message hard-coded in the class.

BTW if anyone asks you: How many classes does a Java enterprise developer need to print “Hello world!" you can proudly reply: 4 classes, 30 JARs taking 20 MiB of space and a relational database with 10+ tables. Seriously, this is an output of our article here...

 

 

Quartz scheduler plugins - hidden treasure

Although briefly described in the official documentation, I believe Quartz plugins aren't known enough, looking at how useful they are.

Essentially plugins in Quartz are convenient classes wrapping registration of underlying listeners. You are free to write your own plugins but we will focus on existing ones shipped with Quartz.

LoggingTriggerHistoryPlugin

First some background. Two main abstractions in Quartz are jobs and triggers. Job is a piece of code that we would like to schedule. Trigger instructs the scheduler when this code should run. CRON (e.g. run every Friday between 9 AM and 5 PM until November) and simple (run 100 times every 2 hours) triggers are most commonly used. You associate any number of triggers to a single job.

Believe it or not, Quartz by default provides no logging or monitoring whatsoever of executed jobs and triggers. There is an API, but no built-in logging is implemented. It won't show you that it now executes this particular job due to this trigger firing. So the first thing you should do is adding the following lines to your quartz.properties:

?

1 2 3 4 5 6 7

org.quartz.plugin.triggerHistory.class=org.quartz.plugins.history.LoggingTriggerHistoryPlugin    org.quartz.plugin.triggerHistory.triggerFiredMessage=Trigger [{1}.{0}] fired job [{6}.{5}] scheduled at: {2, date, dd-MM-yyyy HH:mm:ss.SSS}, next scheduled at: {3, date, dd-MM-yyyy HH:mm:ss.SSS}    org.quartz.plugin.triggerHistory.triggerCompleteMessage=Trigger [{1}.{0}] completed firing job [{6}.{5}] with resulting trigger instruction code: {9}. Next scheduled at: {3, date, dd-MM-yyyy HH:mm:ss.SSS}    org.quartz.plugin.triggerHistory.triggerMisfiredMessage=Trigger [{1}.{0}] misfired job [{6}.{5}]. Should have fired at: {3, date, dd-MM-yyyy HH:mm:ss.SSS}

The first line (and the only required) loads the plugin class LoggingTriggerHistoryPlugin. The remaining lines are configuring the plugin, customizing the logging messages. I found the built-in defaults not very well thought, e.g. they display current time which is already part of the logging framework message. You are free to construct any logging message, see the API for details. Adding these extra few lines makes debugging and monitoring much easier:

?

1 2 3

LoggingTriggerHistoryPlugin | Trigger [Demo.Every-few-seconds] fired job [Demo.Print-message] scheduled at:  04-04-2012 23:23:47.036, next scheduled at:  04-04-2012 23:23:51.036 //...job output LoggingTriggerHistoryPlugin | Trigger [Demo.Every-few-seconds] completed firing job [Demo.Print-message] with resulting trigger instruction code: DO NOTHING. Next scheduled at:  04-04-2012 23:23:51.036

You see now why naming your triggers (Demo.Every-few-seconds) and jobs (Demo.Print-message) is so important.

LoggingJobHistoryPlugin

There is another handy plugin related to logging:

?

1 2 3 4 5

org.quartz.plugin.jobHistory.class=org.quartz.plugins.history.LoggingJobHistoryPlugin org.quartz.plugin.jobHistory.jobToBeFiredMessage=Job [{1}.{0}] to be fired by trigger [{4}.{3}], re-fire: {7} org.quartz.plugin.jobHistory.jobSuccessMessage=Job [{1}.{0}] execution complete and reports: {8} org.quartz.plugin.jobHistory.jobFailedMessage=Job [{1}.{0}] execution failed with exception: {8} org.quartz.plugin.jobHistory.jobWasVetoedMessage=Job [{1}.{0}] was vetoed. It was to be fired by trigger [{4}.{3}] at: {2, date, dd-MM-yyyy HH:mm:ss.SSS}

The rule is the same - plugin + extra configuration. See JavaDoc of LoggingJobHistoryPlugin for details and possible placeholders. Quick look at logs reveals very descriptive output:

?

1 2 3 4 5

Trigger [Demo.Every-few-seconds] fired job [Demo.Print-message] scheduled at:  04-04-2012 23:34:53.739, next scheduled at:  04-04-2012 23:34:57.739 Job [Demo.Print-message] to be fired by trigger [Demo.Every-few-seconds], re-fire: 0 //...job output Job [Demo.Print-message] execution complete and reports: null Trigger [Demo.Every-few-seconds] completed firing job [Demo.Print-message] with resulting trigger instruction code: DO NOTHING. Next scheduled at:  04-04-2012 23:34:57.739

I have no idea why these plugins aren't enabled by default. After all, if you don't want such a verbose output, you can turn it off in your logging framework. Never mind, I think it is a good idea to have them in place when troubleshooting Quartz execution.

XMLSchedulingDataProcessorPlugin

This is a pretty comprehensive plugin. It reads XML file (by default named quartz_data.xml) containing jobs and triggers definitions and adds them to the scheduler. This is especially useful when you have a global job that you need to add once. Plugin can either update the existing jobs/triggers or ignore the XML file if they already exist - very useful when JDBCJobStore is used.

?

1	org.quartz.plugin.xmlScheduling.class=org.quartz.plugins.xml.XMLSchedulingDataProcessorPlugin

In the aforementioned article we have been manually adding job to the scheduler:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

val trigger = newTrigger().         withIdentity("Every-few-seconds", "Demo").         withSchedule(             simpleSchedule().                     withIntervalInSeconds(4).                     repeatForever()         ).         build()     val job = newJob(classOf[PrintMessageJob]).         withIdentity("Print-message", "Demo").         usingJobData("msg", "Hello, world!").         build()     scheduler.scheduleJob(job, trigger)

The same can be achieved with XML configuration, just place the following quartz_data.xml in your CLASSPATH:

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38

<?xml version="1.0" encoding="UTF-8"?> <job-scheduling-data xmlns="http://www.quartz-scheduler.org/xml/JobSchedulingData"                      xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"                      xsi:schemaLocation=" http://www.quartz-scheduler.org/xml/JobSchedulingData http://www.quartz-scheduler.org/xml/job_scheduling_data_2_0.xsd ">        <processing-directives>         <overwrite-existing-data>false</overwrite-existing-data>         <ignore-duplicates>true</ignore-duplicates>     </processing-directives>        <schedule>         <trigger>             <simple>                 <name>Every-few-seconds</name>                 <group>Demo</group>                 <job-name>Print-message</job-name>                 <job-group>Demo</job-group>                 <repeat-count>-1</repeat-count>                 <repeat-interval>4000</repeat-interval>             </simple>         </trigger>            <job>             <name>Print-message</name>             <group>Demo</group>             <job-class>com.blogspot.nurkiewicz.quartz.demo.PrintMessageJob</job-class>             <job-data-map>                 <entry>                     <key>msg</key>                     <value>Hello, World!</value>                 </entry>             </job-data-map>         </job>        </schedule>       </job-scheduling-data>

The file supports both simple and CRON triggers and is well described using XML Schema. It is even possible to point out to an XML files somewhere in the file system and periodically scan them for changes (!) (see: XMLSchedulingDataProcessorPlugin.setScanInterval(). Guess what is Quartz using to schedule periodic scanning?

?

1 2	org.quartz.plugin.xmlScheduling.fileNames=/etc/quartz/system-jobs.xml,/home/johnny/my-jobs.xml org.quartz.plugin.xmlScheduling.scanInterval=60

ShutdownHookPlugin

Last but not least, ShutdownHookPlugin. Small but probably useful plugin that register shutdown hook in the JVM in order to gently stop the scheduler. However I recommend turning cleanShutdown off - if the system already tries to abruptly stop the application (typically scheduler shutdown is called by Spring via SchedulerFactoryBean) or the user hit Ctrl+C - waiting for currently running jobs seems like a bad idea. After all, maybe we are killing the application because some jobs are running for too long/hunging?

?

1 2	org.quartz.plugin.shutdownHook.class=org.quartz.plugins.management.ShutdownHookPlugin org.quartz.plugin.shutdownHook.cleanShutdown=false

As you can see Qurtz ships with few quite interesting plugins. For some reason they aren't described in detail in the official documentation, but they work pretty well and are a valuable addition to scheduler.

The source code with applied plugins is available on GitHub.