Manage execution order

The wait_for argument defines the dependencies of a step within the pipeline.

If wait_for=None, the step has no dependencies and will start executing immediately.
If a step does not explicitly set wait_for, it will automatically depend on all preceding steps in the pipeline.

Code

from lightning_sdk.pipeline import Pipeline, JobStep
from lightning_sdk.machine import Machine

pipeline = Pipeline(name='second-pipeline')
pipeline.run(
  steps=[
      JobStep(
        name='job-1',
        image="ubuntu:latest",
        machine=Machine.CPU,
        command="sleep 9 && echo 'Hello, World!'"
      ),
      JobStep(
          name='job-2',
          image="ubuntu:latest",
          machine=Machine.CPU,
          command="sleep 9 && echo 'Hello, World!'",
          wait_for=None
      ),
      JobStep(
          name='job-3',
          image="ubuntu:latest",
          machine=Machine.CPU,
          command="sleep 9 && echo 'Hello, World!'",
      ),
  ]
)

The resulting Pipeline

===== Generated Pipeline =====
0 - Job['job-1'] wait_for nothing
1 - Job['job-2'] wait_for nothing
2 - Job['job-3'] wait_for ['job-1', 'job-2']

Was this page helpful?