With Resume

The --resume flag tells dotflow start to skip tasks that already have a checkpoint in the configured storage.

Usage

dotflow start --workflow docs_src.cli.cli_with_workflow:pipeline --resume

When to use

• A previous run failed mid-pipeline and you fixed the root cause
• A long-running pipeline was killed by infra (Lambda timeout, OOM, instance preemption)
• You want to verify a specific step in isolation without re-running upstream

Requirements

• A fixed workflow_id so dotflow can locate the previous checkpoints
• A persistent storage provider (StorageFile, StorageS3, StorageGCS)

Set WORKFLOW_ID via environment, or expose it inside the workflow factory.

WORKFLOW_ID=12345678-1234-5678-1234-567812345678 \
  dotflow start --workflow docs_src.cli.cli_with_workflow:pipeline --resume

Combine with --step

--resume works with a single-step entry as well:

dotflow start --step docs_src.cli.cli_with_workflow:step_one --resume

Input change

If the initial_context differs from the stored checkpoint, the behavior is controlled by on_input_change (see Checkpoint with input-change policy).

References

• Checkpoint
• With Workflow