Skip to content

Observability Operations

This guide covers production observability operations for Container Apps using Log Analytics, Application Insights, and distributed tracing.

Signals and Alerting Architecture

flowchart TD
    METRICS[Metrics: Requests, ResponseTime, CPU, Memory, Replicas] --> METRIC_ALERTS[Metric alert rules]
    LOGS[Log Analytics queries:<br/>ContainerAppConsoleLogs_CL<br/>ContainerAppSystemLogs_CL] --> LOG_ALERTS[Log search alert rules]
    METRIC_ALERTS --> AG[Action Group]
    LOG_ALERTS --> AG
    AG --> EMAIL[Email]
    AG --> TEAMS[Teams]
    AG --> WEBHOOK[Webhook]

Prerequisites

  • Log Analytics workspace connected to the Container Apps environment
  • Application Insights configured for application telemetry
export RG="rg-myapp"
export APP_NAME="ca-myapp"
export ENVIRONMENT_NAME="cae-myapp"

Log Analytics Operations

Identify workspace connected to the environment:

az containerapp env show \
  --name "$ENVIRONMENT_NAME" \
  --resource-group "$RG" \
  --query "properties.appLogsConfiguration" \
  --output json

Example output (PII scrubbed):

{
  "destination": "log-analytics",
  "logAnalyticsConfiguration": {
    "customerId": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
  }
}

Run a KQL query for recent errors:

az monitor log-analytics query \
  --workspace "<log-analytics-workspace-id>" \
  --analytics-query "ContainerAppConsoleLogs_CL | where ContainerAppName_s == '$APP_NAME' | where Log_s contains 'ERROR' | limit 50" \
  --output table

Example output:

No results found.

Application Insights Operations

List availability and request telemetry for the app:

az monitor app-insights query \
  --app "<app-insights-name>" \
  --resource-group "$RG" \
  --analytics-query "requests | where cloud_RoleName == '$APP_NAME' | summarize count() by resultCode, bin(timestamp, 5m)" \
  --output table

Use container logs directly during active incidents:

az containerapp logs show \
  --name "$APP_NAME" \
  --resource-group "$RG" \
  --type console \
  --follow false

Example output from the running revision:

{"status":"healthy","timestamp":"2026-04-04T11:32:37.322216+00:00"}

Track replica and revision health as platform signals:

az containerapp revision list \
  --name "$APP_NAME" \
  --resource-group "$RG" \
  --output json

az containerapp replica list \
  --name "$APP_NAME" \
  --resource-group "$RG" \
  --revision "ca-myapp--0000001" \
  --output json

Example output (PII scrubbed):

[
  {
    "name": "ca-myapp--0000001",
    "active": true,
    "trafficWeight": 100,
    "replicas": 1,
    "healthState": "Healthy",
    "runningState": "Running"
  }
]

[
  {
    "name": "ca-myapp--0000001-646779b4c5-bhc2v",
    "properties": {
      "containers": [
        {
          "name": "ca-myapp",
          "ready": true,
          "restartCount": 0,
          "runningState": "Running"
        }
      ],
      "runningState": "Running"
    }
  }
]

Distributed Tracing Operations

Confirm trace context propagation across services by querying end-to-end operation IDs in Application Insights.

az monitor app-insights query \
  --app "<app-insights-name>" \
  --resource-group "$RG" \
  --analytics-query "dependencies | where cloud_RoleName == '$APP_NAME' | project timestamp, operation_Id, target, resultCode | limit 20" \
  --output table

Verification Steps

Check that logs and traces are flowing within expected delay windows.

az monitor app-insights component show \
  --app "<app-insights-name>" \
  --resource-group "$RG" \
  --output json

Example output (PII masked):

{
  "id": "/subscriptions/<subscription-id>/resourceGroups/rg-myapp/providers/microsoft.insights/components/<app-insights-name>",
  "name": "<app-insights-name>",
  "provisioningState": "Succeeded"
}

Observability Decision Matrix

Signal Type Best For Query Surface Typical Alert Latency
Platform metrics Fast saturation and availability detection Azure Monitor metrics 1-2 minutes
Console/system logs Detailed failure context and root cause hints Log Analytics (KQL) 2-5 minutes
Distributed traces Cross-service request path analysis Application Insights 2-5 minutes

Pair every alert with an investigation query

For each alert rule, store a companion KQL query that responders can run immediately. This shortens MTTR by removing first-response guesswork.

Avoid unbounded log volume

Excessive debug logging can increase ingestion cost and hide actionable events. Use structured JSON logs and severity controls in production.

Telemetry Freshness Workflow

sequenceDiagram
    participant App as Container App
    participant LA as Log Analytics
    participant AI as Application Insights
    participant Ops as On-call Engineer
    App->>LA: Console and system logs
    App->>AI: Traces and requests
    Ops->>LA: KQL error and restart checks
    Ops->>AI: End-to-end trace validation
    Ops->>Ops: Decide rollback or mitigation

Troubleshooting

No logs in workspace

  • Confirm environment log configuration points to the expected workspace.
  • Check regional alignment between app, environment, and workspace.
  • Validate IAM permissions for querying telemetry resources.

Missing distributed traces

  • Verify OpenTelemetry exporter endpoint and connection string settings.
  • Ensure incoming requests include trace context headers.

Advanced Topics

  • Define SLO-based alerts (latency, error rate, saturation).
  • Build dashboards combining infra metrics and app traces.
  • Use sampling strategies to balance fidelity and telemetry cost.

See Also

Sources