debug-deployment

Run dlthub job runs info with the job name to check exit status and timing, then use dlthub job runs logs to read the error output. Common causes include missing dependencies in pyproject.toml, unconfigured prod profile secrets, or leftover dev_mode settings.

Use dlthub job logs with a job name or selector for the latest run, or dlthub job runs logs with a run number for a specific execution. Add the -f flag to stream logs in real time.

The most common cause is a script missing the if __name__ == "__main__": guard, so the job executes no code. Also verify all dependencies are declared in pyproject.toml, since locally installed packages are not automatically available at runtime.

Yes, but only read-only and after explicit human confirmation. Pin the access or prod profile, run commands with dlthub local run --profile prod or the WORKSPACE__PROFILE environment variable, then pin the dev profile again when finished.

Use dlthub job cancel with a job name or selector such as tag:backfill to cancel active runs. Add --dry-run to preview what would be cancelled, or use dlthub job runs cancel to stop a specific run.

The default job timeout is 120 minutes. Override it by setting execute={"timeout": "6h"} in the job decorator to allow longer-running pipelines to complete.