Feature description

The JDBC Batch Task can fail due to transient issues while processing input files (e.g. interrupted file input streams, temporary I/O or network glitches). While we already have ongoing work to mitigate some of these issues (see https://github.com/kestra-io/kestra-ee/issues/6338), the Batch Task itself could be more resilient.

If a transient failure occurs during batch processing, the task fails and requires a full retry from the beginning, even though the failure may be recoverable.

Proposed improvement

Enhance the JDBC Batch Task to:

• Automatically retry on transient failures (e.g. I/O interruptions, temporary connectivity issues)
• On retry, refetch the full input file and resume processing from the last successful iteration
• Avoid retries on non-transient errors, such as database constraint violations or conflicts
• Make retry behavior configurable with a maximum retry count and sensible default

See also the original Slack thread