Skip to main content

Schedules and job flows

A workflow defines how a piece of work happens. A job defines how workflow executions are organised and when they run. Use jobs for recurring work, multi-step processes, and worker-based batch processing.

For example, a job flow could prepare a set of documents, process them with a batch step, then run a final workflow to summarise the results.

Put a workflow into a job​

  1. Create and test the workflows you need in the project.
  2. Open Jobs, create a job, and open its flow editor.
  3. Add the required steps, choose the workflow for each step, and configure its inputs.
  4. Connect the steps in their intended order and review the saved configuration.
  5. Use Run now to test the whole job flow before adding a recurring schedule.

The job flow editor coordinates workflow executions. It is separate from the workflow's node editor, where you build the logic inside each step.

Single-run and batch steps​

A single-run step executes its selected workflow once. Use it for setup, an individual task, or a final summary.

A batch step uses a worker model: the number of batch runs is the number of workers assigned to that step. Inside the workflow, use Query Batch Data nodes to retrieve and iterate through the data assigned to each worker. Worker count is not the number of records being processed, and does not guarantee that all workers execute simultaneously.

A job can combine single-run and batch steps. Think of batching as a property of a step in the flow, not as a separate kind of job.

Worker count and pod concurrency are different controls: the first determines how many batch runs the step creates, while the second limits how many workflow runs a runtime pod can process at once. Workflows assigned to the same runtime group share that capacity with its other work.

Pass data between steps​

Use named Input nodes for the values a step receives and Set Output nodes for results needed by later steps. Configure the job step's input mapping to use the relevant values from the preceding step.

Connecting job cards establishes their sequence; it does not automatically expose every intermediate node value inside the workflows. Return the values the next step needs explicitly.

Add a schedule​

In the job's Run options, choose Schedule, configure the recurrence and timezone, and click Save schedule. Review the entry under Scheduled runs. The schedule runs the job's complete flow, not just the selected step.

You can also use Run after delay for a one-off queued run. Queued work may wait for execution capacity, so do not assume the requested start time guarantees immediate execution.

Before leaving a job unattended, test its inputs, external connections, outputs, and failure behaviour. Consider whether repeating a run could duplicate writes or notifications.

Monitor and stop work​

Use the job's Runs table to check status, step progress, timing, and messages. Open Logs on a run to inspect that execution, including its workflow-node messages.

Removing a schedule stops its future runs and cancels its queued runs; it does not stop an execution that has already started. Cancel that run separately if needed. Cancellation does not undo completed external actions.

The Jobs guide covers the flow editor, worker configuration, schedules, run table, and logs in detail, with screenshots.