Workflows using Job Chains

JobRunr’s job chaining and workflow capabilities enhance software development efficiency by enabling logical task sequencing and complex workflow management. This setup ensures tasks execute in order, maintaining data integrity and operational consistency, especially when subsequent processes depend on earlier outcomes. The Pro version offers advanced controls and insights, further optimizing performance and providing you with an instant overview of your business process.

This comes in really handy when using Batches - start a new step in your business process when a whole bunch of related jobs have finished.

Job chaining via continueWith and onFailure

Running a job after another job succeeded

@Inject
private ArchiveService archiveService;
@Inject
private NotifyService notifyService;

public void createArchiveAndNotify(String folder) {
    BackgroundJob
        .enqueue(() -> archiveService.createArchive(folder))
        .continueWith(() -> notifyService.notifyViaSlack("ops-team", "The following folder was archived: " + folder));
}

Do you have other tasks to run after the slack notification is sent? You can keep on chaining jobs using continueWith. BUT if at any step of the chain, a job fails, JobRunr will not execute the remaining tasks in the chain.

Running a job after another job failed

@Inject
private ArchiveService archiveService;
@Inject
private NotifyService notifyService;

public void createArchiveAndNotify(String folder) {
    BackgroundJob
        .enqueue(() -> archiveService.createArchive(folder))
        .onFailure(() -> notifyService.notifyViaSlack("ops-team", "The following folder could NOT be archived: " + folder));
}

Contrary to continueWith, onFailure does not allow for the chain to continue. If you try, the editor will complain as the method returns void. Essentially, onFailure indicates the end of the chain.

Running a job after another job succeeded or failed

@Inject
private ArchiveService archiveService;
@Inject
private NotifyService notifyService;

public void createArchiveAndNotify(String folder) {
    BackgroundJob
        .enqueue(() -> archiveService.createArchive(folder))
        .continueWith(
            () -> notifyService.notifyViaSlack("ops-team", "The following folder was archived: " + folder), // on success
            () -> notifyService.notifyViaSlack("ops-team", "The following folder could NOT be archived: " + folder) // on failure
        );
}

Yes, here too you can keep the chain going using continueWith! If the previous job succeeds, the on success job is executed. BUT if at any step of the chain, a job fails, JobRunr will end the chain with the execution of the on failure task. Essentially, the execution of a on failure job always ends the chain, while execution of a on success job keeps the chain going.

Job chaining when using the JobBuilder

If you are using the JobBuilder pattern, this is also possible by means of the runAfter method.

@Inject
private ArchiveService archiveService;
@Inject
private NotifyService notifyService;

public void createArchiveAndNotify(String folder) {
    JobId createArchiveJobId = BackgroundJob
        .create(aJob()
            .withDetails(() -> archiveService.createArchive(folder)));

    JobId notifyViaSlackJobId = BackgroundJob
        .create(aJob()
            .runAfter(createArchiveJobId)
            .withDetails(() -> notifyService.notifyViaSlack("ops-team", "The following folder was archived: " + folder)));
}

How does it work?

• the first job (archiveService.createArchive(folder)) is enqueued and will start processing as soon as some worker threads are available
• the second job (notifyService.notifyViaSlack(String room, String message)) will initially be saved using the AWAITING state.
• once the first job either succeeds or fails, JobRunr will look for the relevant job to enqueue and process.

Note: you shouldn’t expect notifyViaSlack to be the exact first job executed by JobRunr after createArchive. The reason being that JobRunr runs tasks in the order they are ENQUEUED (unless priorities are involved).