Worker Configuration

Workers can be created via the hatchet.worker() method, after instantiating a hatchet instance. The hatchet.worker() method takes the following arguments:

• name (required): The name of the worker. This is used to identify the worker in the Hatchet UI.
• max_runs: The maximum number of concurrent step runs that the worker can run. If not set, it defaults to 100. Note that this value is different from the number of concurrent runs per workflow.

Registering Workflows

Workers can register workflows by calling worker.register_workflow with a workflow class instance. There is no limit to the number of workflows which can be registered for each worker. For example:

from hatchet_sdk import Context, Hatchet
 
hatchet = Hatchet(debug=True)
 
@hatchet.workflow()
class WorkflowOne:
    @hatchet.step()
    async def step(self, context: Context):
        return {"test1": "test1"}
 
@hatchet.workflow()
class WorkflowTwo:
    @hatchet.step()
    async def step(self, context: Context):
        return {"test2": "test2"}
 
worker = hatchet.worker('test-worker')
worker.register_workflow(WorkflowOne())
worker.register_workflow(WorkflowTwo())
worker.start()

Starting a Worker

Workers can be started by calling either worker.start or worker.async_start. We recommend that worker.start is the last call made when running a worker.

The worker.start method is blocking, while worker.async_start can be awaited or started via asyncio.create_task.

Example: worker.start

from hatchet_sdk import Context, Hatchet
 
hatchet = Hatchet()
 
# ... workflow definitions
 
def main():
    worker = hatchet.worker("test-worker", max_runs=4)
    worker.register_workflow(Workflow())
    worker.start()
 
main()

Example: worker.async_start

from hatchet_sdk import Context, Hatchet
 
hatchet = Hatchet()
 
# ... workflow definitions
 
async def main():
    worker = hatchet.worker("test-worker", max_runs=4)
    worker.register_workflow(Workflow())
    await worker.async_start()
 
asyncio.run(main())