Make the producer sign for the nothing it produced

Description and use

In automated pipelines, an empty queue often masquerades as a successful run. By having the producer sign its output, you can explicitly distinguish a legitimate idle run from a silent failure, turning invisible gaps into actionable incidents.

Problem

Unattended automation pipelines report success even when no work was performed, because the producer silently writes nothing to the queue, making it impossible to distinguish between a legitimate empty run and a silent failure.

Outcome

By requiring the producer to sign its output, the pipeline gains explicit visibility into why a queue is empty, turning silent gaps into actionable incidents and improving reliability of unattended automation.

Workflow

• 1. Producer scans for work items and records the count of examined items, rejected items, and the reason for zero production if applicable.
• 2. Producer writes a signed record (e.g., a log entry or metadata file) describing its findings.
• 3. Consumer opens the queue and processes any items present.
• 4. Consumer checks for the producer’s signature from the current cycle.
• 5. If the signature indicates zero items with a valid reason, consumer exits cleanly; if the signature is missing, stale, or from a previous cycle, consumer marks the run as an incident.
• 6. The pipeline’s summary reflects the explicit state (success, incident, or failure).

Evidence

The producer is no longer allowed to communicate by absence. It has to sign for its own output, including when that output is nothing.

Source

Signature present and it says zero examined for a legible reason, fine, exit clean. Signature absent, or stale, or from yesterday, and the run is not a success — it is a run that could not establish whether it had work.

Source