Method execute

axlearn/cloud/common/job.py:115–132

Wraps _execute with retries. All args and kwargs are forwarded. Retries are triggered automatically when _execute throws an exception. If all retries are exhausted, _delete is invoked to cleanup the job.

(self)

    # TODO(markblee): Expand the API to include create/delete/start/stop.
    def execute(self) -> Any:
        """Wraps _execute with retries. All args and kwargs are forwarded.

        Retries are triggered automatically when _execute throws an exception.
        If all retries are exhausted, _delete is invoked to cleanup the job.
        """
        cfg = self.config

        try:
            return _with_retry(
                self._execute,
                interval=cfg.retry_interval,
                max_tries=cfg.max_tries,
            )
        except Exception:  # pylint: disable=broad-except
            # Cleanup on unexpected failure or exhausted retries.
            self._delete()
            raise


# TODO(markblee): Consider adding exponential backoff.