SIGUSR1 进程内重启后执行状态重置排查

• Gateway supports in-process restart flow and you can access runtime logs.
• You have a reproducible exec task that previously got stuck after restart.

1. Start one long-running exec/process task and record its session id/state before restart.
2. Trigger in-process restart and watch logs for stale execution state reset handling.
3. After restart, poll the old session id and ensure no ghost-running state remains.
4. Run a fresh exec task to validate scheduler/worker path is healthy.

• If restart happens during critical writes, still keep idempotency/compensation logic in application layer.
• PR detail page may require signed-in context for full diff review（需验证）。

This tip is aggregated from community/public sources and preserved with attribution.