A time-based anomaly is an irregularity in event timing, ordering, or correlation that departs from expected system behavior. Delays, out-of-order operations, duplicate retries, and mismatched timestamps can expose coordination or consistency problems without immediate exceptions. section 13.1.1
Investigate the anomaly across the complete transaction path:
- Follow requests with distributed tracing tools such as OpenTelemetry or Zipkin; treat missing spans and trace trees ending before writes or publishing as red flags. section 13.1.1
- Compare application logs with Kafka topic inspection to confirm whether messages propagated through the system. section 13.1.1
- Query the database for missing records or sequence gaps, then inspect PostgreSQL deadlock logs for concurrency failures and unhandled retries. section 13.1.1
Payment trace that stops early
A payment trace reaches the payment-processing service but records no further spans. Checking POS logs, the Kafka topic, and database evidence can identify whether auditing, message propagation, or transaction handling failed.
