Microsoft Docs

Monitoring Azure Functions

Effective monitoring is crucial for understanding the health, performance, and behavior of your Azure Functions. This document outlines the key tools and techniques for monitoring your serverless applications.

Application Insights Integration

Azure Functions integrates seamlessly with Application Insights, a powerful application performance management (APM) service. Application Insights provides detailed telemetry about your function executions, including:

• Requests: Track incoming triggers and their success/failure rates.
• Dependencies: Monitor calls to external services (databases, HTTP endpoints, etc.).
• Exceptions: Log and analyze runtime errors.
• Traces: Capture custom log messages from your function code.
• Performance Metrics: Analyze execution duration, CPU usage, memory consumption, and more.

To enable Application Insights for your Azure Function App, ensure it's configured in your function app's settings. You can access the Application Insights portal directly from your Function App's overview page in the Azure portal.

Live Metrics Stream

The Live Metrics Stream in Application Insights provides near real-time monitoring of your function app. This is invaluable for observing immediate impacts of deployments or identifying sudden issues.

Kusto Query Language (KQL) for Advanced Analysis

Application Insights logs are stored in Azure Monitor Logs, which you can query using Kusto Query Language (KQL). This allows for deep analysis of your telemetry data.

Here's a simple KQL query to find the 10 slowest function executions in the last hour:

traces
| where timestamp > ago(1h)
| where message contains "Function completed successfully"
| extend duration = toint(tostring(customDimensions.DurationMs))
| order by duration desc
| take 10
            

Azure Monitor Alerts

Set up alerts in Azure Monitor to be proactively notified of critical events or performance degradations. You can configure alerts based on various metrics and logs, such as:

• High error rates
• Increased function execution duration
• Specific log messages
• Resource utilization

Alerts can be configured to send notifications via email, SMS, or trigger automated actions like running another Azure Function or calling a webhook.

Function App Metrics

Beyond Application Insights, the Azure portal provides built-in metrics for your Function App, offering a high-level view of:

• Function Execution Count
• Function Execution Units
• HTTP 5xx Errors
• Average Execution Time

These metrics can be viewed directly on the Function App's "Metrics" blade and can also be used to configure Azure Monitor alerts.

Logging Best Practices

• Use the built-in logging functions (e.g., context.log in JavaScript/TypeScript, logging.info in Python, ILogger in C#) to record important events, parameters, and outcomes.
• Structure your logs for easier parsing and analysis. Include relevant identifiers like request IDs or correlation IDs.
• Avoid logging sensitive information directly.