Monitoring

The platform provides monitoring across three pillars: metrics, logs, and traces — all backed by Azure-managed services.

Metrics (Prometheus + Grafana)

Azure Monitor Workspace

AKS Automatic ships built-in Prometheus metrics collection via Azure Monitor Workspace. This captures:

• Kubernetes control plane metrics (API server, scheduler, etcd)
• Node-level metrics (CPU, memory, disk, network)
• Pod-level metrics (container resource usage)
• Istio mesh metrics (request volume, latency, error rate)

Grafana

When enabled (ENABLE_GRAFANA_WORKSPACE=true), Azure Managed Grafana provides pre-built dashboards for:

• Cluster health and node utilization
• Pod resource consumption
• Istio service mesh traffic
• Karpenter node provisioning

Tip

Disable Grafana (azd env set ENABLE_GRAFANA_WORKSPACE false) for dev/test environments to reduce costs. Prometheus metrics are still collected.

Logs (Container Insights)

Log Analytics Workspace

Container Insights collects logs from all containers and forwards them to Log Analytics. Query logs via Azure Portal or az monitor log-analytics query:

// Pod logs for a specific service
ContainerLogV2
| where PodNamespace == "osdu"
| where PodName startswith "partition"
| project TimeGenerated, LogMessage
| order by TimeGenerated desc
| take 100

Common Log Queries

// Services in CrashLoopBackOff
KubePodInventory
| where Namespace == "osdu"
| where PodStatus == "Failed"
| summarize count() by Name, PodStatus

// OOM kills
ContainerLogV2
| where LogMessage contains "OOMKilled"
| project TimeGenerated, PodName, LogMessage

Traces (Application Insights)

OSDU services emit distributed traces to Application Insights via the APPLICATIONINSIGHTS_CONNECTION_STRING environment variable. This enables:

• End-to-end request tracing across services
• Dependency maps showing service-to-service and service-to-PaaS calls
• Exception tracking and failure analysis
• Performance analysis (latency percentiles, throughput)

Middleware Monitoring

Elasticsearch

Kibana is optionally exposed via the gateway module. Access the Kibana dashboard to monitor:

• Cluster health (green/yellow/red)
• Index status and document counts
• Search query performance
• Shard allocation across nodes

Airflow

The Airflow web UI is optionally exposed via the gateway module. Monitor:

• DAG execution status
• Task run history and durations
• Worker pod scheduling