Monitoring Azure Functions

This tutorial guides you through the essential aspects of monitoring your Azure Functions applications to ensure they are healthy, performant, and reliable.

Introduction

Effective monitoring is crucial for understanding the behavior of your serverless applications, diagnosing issues, and optimizing performance. Azure Functions integrates deeply with Azure's monitoring services, providing comprehensive insights into your function executions.

Key Monitoring Tools

Azure offers a suite of powerful tools for monitoring your functions:

Application Insights: A powerful Application Performance Management (APM) service that provides deep insights into your application's performance and usage.
Azure Monitor: A unified monitoring solution that collects, analyzes, and acts on telemetry from your Azure and on-premises environments.
Azure Portal: The central hub for viewing logs, metrics, and configuring alerts.

Application Insights

Application Insights is your primary tool for detailed monitoring of Azure Functions. It automatically captures a wealth of telemetry data.

Key Features:

Live Metrics: Real-time monitoring of requests, dependencies, exceptions, and performance counters.
Failures: Identifies and analyzes exceptions, providing stack traces and request context.
Performance: Tracks request duration, averages, and patterns to pinpoint performance bottlenecks.
Dependency Tracking: Monitors calls made to external services (databases, APIs, etc.).
Log Streaming: View function logs in real-time.

Enabling Application Insights:

Application Insights is typically enabled by default when you create an Azure Function App. If not, you can link it manually through the Azure portal.

Ensure your host.json file is configured to send logs to Application Insights if you're not using the default integration.


{
  "version": "2.0",
  "logging": {
    "applicationInsights": {
      "samplingSettings": {
        "isEnabled": true,
        "excludedTypes": "Request"
      }
    }
  }
}

Azure Monitor

Azure Monitor provides a broader view of your Azure resources, including metrics and logs for your Function App.

Metrics:

Function Execution Count: The number of times your functions have run.
Http 5xx: The number of server-side errors.
Memory Working Set: The amount of memory your function app is using.
Requests: The number of HTTP requests received by your function app.

Logs:

You can query logs stored by Application Insights using Kusto Query Language (KQL) in the Azure portal's Log Analytics workspace.

Example KQL query to see function execution logs:


requests
| where success == "false"
| project timestamp, name, url, resultCode, duration, cloud_RoleName, itemType
| order by timestamp desc

Logging Strategies

Effective logging is key to debugging and monitoring.

Log Function Inputs and Outputs: Log key parameters passed to your functions and the results they produce.
Use Different Log Levels: Employ Information, Warning, Error, and Debug levels appropriately.
Correlation IDs: Implement correlation IDs to trace a single request across multiple functions or services.
Structured Logging: Log data in a structured format (like JSON) for easier parsing and analysis.

Example of structured logging:


using Microsoft.Azure.WebJobs;
using Microsoft.Extensions.Logging;
using Newtonsoft.Json;

public static class MyFunction
{
    [FunctionName("MyFunction")]
    public static void Run([QueueTrigger("myqueue-items")] string myQueueItem, ILogger log)
    {
        log.LogInformation($"C# Queue trigger function processed: {myQueueItem}");

        // Simulate some processing
        var data = JsonConvert.DeserializeObject(myQueueItem);
        var correlationId = data?.correlationId ?? Guid.NewGuid().ToString();
        var processingResult = new { status = "success", processedItems = 1 };

        log.LogInformation($"Processing completed. CorrelationId: {correlationId}", new { correlationId, processingResult });
    }
}

Alerting

Set up alerts to be notified proactively when critical events occur.

Common Alerts:

High rate of HTTP 5xx errors.
High function execution duration.
Application Insights availability test failures.
Custom log searches (e.g., specific error messages).

You can configure alerts directly within the Azure portal under the "Alerts" section of your Function App or Log Analytics workspace.

Performance Tuning

Monitor your functions to identify areas for performance improvement.

Optimize Cold Starts: Consider pre-warming strategies or choosing the Premium plan for consistent performance.
Efficient Dependencies: Ensure your code efficiently interacts with external services.
Scaling: Understand your app's scaling behavior and configure appropriate scaling limits if using the Premium plan.

Next Steps

Explore the detailed documentation for Application Insights and Azure Monitor to master advanced querying and alerting techniques.