[mrgoonie]

Summary

Adds one paragraph, in EN and VI, to Schedules and heartbeat → Retries, timeouts and run history. It covers how cron runs behave in the gateway after nextlevelbuilder/dewee#1467 (merged) and nextlevelbuilder/dewee#1534:

• A job runs at most once at a time. Run now or dewee cron run on a running job is refused with "schedule is already running". A scheduled time that comes up during a run is skipped, not queued.
• A run cut short by a crash or restart is marked interrupted and the job is scheduled again. On PostgreSQL that happens about two minutes after the run stops heartbeating. On Lite it happens when the gateway starts.
• A run on another gateway instance keeps going when the scheduler moves between instances.

The source of truth is docs/08-scheduling-cron.md in the gateway repo, section "Run claims and interrupted runs".

Held as draft until nextlevelbuilder/dewee#1534 merges into the gateway's dev, so the docs do not describe behavior that has not landed.

Checks

• pnpm check: 0 errors
• pnpm test: 35 files, 466 tests passed
• pnpm build: complete

Refs nextlevelbuilder/dewee#1510, nextlevelbuilder/dewee#1515

---

Generated by Claude Code