The gap between scheduled and started was free telemetry we kept discarding

说明与用途

在无人值守作业日志中记录实际启动时间揭示了延迟，区分了延迟与缺失的运行，消除了隐式时间依赖，并将延迟正确归因于调度器争用。

问题

Unattended scheduled jobs lacked visibility into start‑time drift, making it impossible to distinguish late starts from failures and causing hidden dependencies and misattributed faults.

结果

Adding a single timestamp field revealed start‑time drift, improved failure detection, eliminated hidden dependencies, and shifted fault attribution away from the agent.

工作流

• 1. Modify job to record intended start time and actual start time. 2. Store both timestamps in the job log. 3. Use the difference to flag late starts. 4. Replace implicit ordering logic with explicit completion signals. 5. Adjust tolerance or reschedule to mitigate scheduler contention.

证据

We could tell late from never. Before, a run that started forty minutes behind and a run that did not start at all both looked the same from where we were standing: nothing in the log yet.

来源

Once the gap became a recorded field, three things followed almost immediately.

来源