Skip to main content

Job configuration

In briefA job is one unit of work in a workflow. It names a job type and its parameters, where it runs, when it builds, what it waits for, and what it does on each outcome. Each job also has an id of its own, given the first time it is saved, which survives a rename.

WhereIn Design › Workflows, open a workflow and open the job in the Job Editor (open its context menu on the canvas › Edit Job). Its tabs are Job Definition, Properties, Frequencies, Dependencies, Events and Documentation. Job Definition holds the Details (including Disabled and Multi-Instance), Job Type, Agent Assignment and Job Parameters sections. The workflow's Code view shows the same job as JSON, including its id.

WhenUse this page when you need to:

  • know what each part of a job controls, and its default;
  • choose how a job is placed on an agent;
  • find out why a version will not commit, or why a job did not build;
  • edit a job's configuration by hand and keep its identity.

How

  1. Name the job. The name must be unique in the workflow and cannot be blank.
  2. Choose the job type and fill its required parameters.
  3. Choose an Agent Assignment mode where the job type needs one: Specific Agent, Pool - Least Tasked or Pool - Run on All.
  4. Add frequencies, dependencies, events and properties.
  5. Save, then commit the workflow to a version.

For the full steps, see Add jobs to a workflow.

Notes
  • A disabled job is left out of every build. Its dependents treat it like a job whose frequency did not fire.
  • On Specific Agent, the secondary, tertiary and quaternary agents are used, in that order, when the primary cannot take the job — see Backup agents.
  • For a legacy job, connectionRefs is what decides the account it runs as. The batch user id shown in the parameters is only a copy of it.
  • An incomplete job blocks Commit changes to a version and is flagged on the canvas.
  • The Documentation tab shows web addresses as links that open in a new tab. Edit opens the text box, and Done returns to the view.
Watch out
  • Leave a job's id alone in the Code view. If you delete it, the job is treated as new.
  • Pasting an id that another workflow already owns is refused at save.
  • A deployment transformation rule cannot target the Connection parameter (batchUserId). Target connectionRefs instead.
  • A disabled job still needs its frequency and calendar references to resolve.
  • Multi-Instance cannot be turned off while the job has more than one property group. Remove all but one first, or in the same save. A job already saved that way fails the build and Add Job.