Verify the evidence channel itself

These examples and illustrative results are independently authored teaching materials, not measured model results.

Use case

After adding import-failure logs/alerts, verify collection, aggregation, lookup and test notification. Teaching test-r9 creates a controlled failure in isolation; logger code alone does not establish a working chain.

Mechanism

Within authorized test scope, generate a unique signal with time/environment. Check structured local fields, aggregator requestId lookup, cross-service traces, metric labels, test receipt and runbook link. Bound ingestion waits and locate missing stages. Use approved test channels and reversible settings, restoring thresholds afterward with scope recorded.

Bad example

Logger calls exist, so monitoring is ready. Missing events mean no failure; ignore collection and alerts.

Good example

Trigger isolated test-r9 import failure and inspect structured logs, aggregation, trace and test alerts by ID. Record found/missing stages and bounded waits, checking destination/runbook. Without execution mark unverified; do not send production alerts.

Why the change matters

Telemetry can fail too. A known input tests the channel, and ID tracing separates absent business activity from lost evidence. Configuration presence does not replace observing outputs.

Observable expectation

Local test-r9 with no aggregate record reveals collection failure, not no error. Wrong-channel notification fails too. A passing teaching record binds event/trace/metric/test receipt to one identity and confirms restored thresholds. This article fires no actual alerts.

Limits

One signal does not establish load or long-term reliability; sampling/redaction changes evidence. Induced errors and notifications follow actual authorization. Disclose unsupported evidence types and restore temporary thresholds; production checks remain separate.