Kubernetes Logs Need More Than Matching Timestamps
Red Hat's log normalizer can give mixed-format logs a common shape. Proving that an event belongs to a request still takes trace context and an end-to-end test.
// Topic
Observability
The Mesh Knows What Your App Does Not
A CNCF reference stack for OpenTelemetry and service-mesh metrics is a practical reminder that application telemetry and proxy telemetry answer different questions. Production observability needs both layers.
Trace Storage Is a Database Problem Again
Jaeger's new ClickHouse backend is a useful reminder that observability cost is not only about how many spans you keep. It is about which questions your storage layout can answer without turning every search into a scan.
End of list