Operational evidence
Real troubleshooting stories.
Each page below turns a runbook into a presentation-friendly narrative: symptom, diagnosis, evidence, and durable fix.
Local Grafana access path. Troubleshooting starts from concrete operator actions, not only abstract failure summaries.
Helm-delivered application exposed through Kubernetes port-forwarding. This is the shortest path to prove the runtime is reachable.
Direct curl verification against the forwarded service. Health and business endpoints become part of the diagnosis evidence.
Scenario library
The point is not to collect random failures. It is to show disciplined debugging on realistic infrastructure, platform, and observability issues.
Helm timeout
Differentiate a slow local startup from a genuinely blocked release.
Open scenarioPrometheus query contract
Show that an API validation error can come from the request shape, not the platform itself.
Open scenarioGrafana localhost mismatch
Separate workstation port-forward assumptions from in-cluster service resolution.
Open scenarioGrafana PVC permissions
Turn a repeated local failure into a cleaner environment design.
Open scenarioNo-data dashboard panel
Prove that empty telemetry can be a traffic and label-filter issue, not a broken stack.
Open scenarioGrafana v2 schema mismatch
Show the difference between a healthy provisioning path and the wrong dashboard export format.
Open scenario