Add Zeta runtime log diagnosis guide
Added bilingual SeaTunnel Zeta documentation for using AI tools to diagnose runtime logs while preserving evidence, redacting sensitive data, and verifying root-cause hypotheses.
apache/seatunnel · #11629
Problem
SeaTunnel users had no official workflow for preparing runtime logs for AI-assisted diagnosis. That made it easy to paste noisy or sensitive logs into tools, accept plausible but unverified explanations, or miss common connector, checkpoint, memory, and retry failure patterns.
Approach
Added English and Chinese log-analysis guides, linked them from the Zeta logging docs, and wired the new page into the documentation sidebar with guidance on log collection, redaction, prompt structure, common failure modes, and a reproducible factory-discovery walkthrough.
Impact and scope
- Gives users a verification-first playbook for diagnosing failed Zeta jobs with AI tools without exposing production secrets.
- Reduces repeated community support loops by documenting how to collect job-scoped logs, preserve first-failure context, and separate evidence from hypotheses.
- Strengthens SeaTunnel's observability documentation with bilingual guidance that connects logs, REST endpoints, connector failures, checkpoints, retries, and escalation steps.
Validation
- Kept the landed branch docs-only after review and confirmed the final diff touched five documentation files.
- Verified formatting, Markdown links, documentation sync, and the English and Chinese Docusaurus builds in the PR.
- Merged after maintainer approval recognized the guide as a solid contribution to the project.