Add local sample dry-run mode for Zeta
Added SeaTunnel Zeta `--dry-run sample` support so local jobs can run configured sources and transforms with bounded samples while replacing real sinks before sink runtime creation.
apache/seatunnel · #11538
Feature impact
SeaTunnel users can preview schemas and bounded sample rows locally without creating the configured sink or writing to target systems.
Problem
SeaTunnel users lacked a local preview mode for understanding what a Zeta job would read and produce. Existing dry-run modes validated configuration and connectivity but did not run source and transform chains with bounded sample output.
Approach
Added CLI parsing and validation for sample dry-run mode, introduced bounded sample configuration, replaced configured sinks with a no-op sample sink, disabled checkpoints, forced deterministic source parallelism, and documented the user-facing command contract in English and Chinese docs.
Impact and scope
- Reduces operational risk by letting users inspect schemas and sample rows before writing to production sinks.
- Keeps sample mode local-only and rejects cluster, async, restore, savepoint, and incompatible execution paths.
- Creates a foundation for richer Zeta debugging workflows while preserving existing static and connect dry-run behavior.
Validation
- Added CLI parsing, validation, sink replacement, bounded collection, and transform-flow sample tests across starter, client, common, core, and server modules.
- Verified focused Maven test suites for command args, job config parsing, source collection, and transform flow lifecycle.
- PR evidence reports package validation across changed engine and starter modules, plus maintainer approval before merge.