Develop Azure Functions with Expert Patterns

This skill covers Azure Functions across multiple programming models: the .NET Isolated Worker Model, the Node.js v4 programming model, and the Python v2 programming model, each with a working template. It covers Durable Functions patterns - function chaining and fan-out/fan-in - plus cold start optimization, queue trigger with error handling, and the HTTP trigger long-running pattern.

It documents eight production sharp edges with fixes: HTTP-triggered functions timing out after 230 seconds regardless of the configured functionTimeout, because Azure's Load Balancer enforces a hard-coded 230-second idle timeout independent of the function app's own setting (fixed with a Durable Functions async pattern, a queue-based async pattern, or a webhook callback - not by raising host.json's timeout, which has no effect on HTTP); socket exhaustion from instantiating a new HttpClient per call (fixed with IHttpClientFactory or a static client, same pattern applying to Azure SDK clients); blocking async calls causing thread starvation (fixed by using async/await throughout rather than synchronous blocking calls); the Consumption plan's 10-minute timeout limit (fixed via Premium plan upgrade, Durable Functions for long workflows, or breaking work into smaller chunks); the .NET in-process model being deprecated with support ending November 10, 2026, after which apps may stop working or receiving security updates (fixed by using the isolated worker model for new projects and migrating existing in-process apps); ILogger not outputting to console or Application Insights (fixed by properly configuring Application Insights and logging levels); missing extension packages causing silent failures (checked via the extension bundle configuration and verifying registered functions in the startup log); and Premium plan instances still experiencing cold starts on scale-out (fixed with a warmup trigger, pre-warmed instance count, or always-ready instances at higher cost).

Use this when building or debugging Azure Functions in .NET, Node.js, or Python - choosing the right programming model, implementing Durable Functions workflows, or diagnosing timeout, cold start, thread-starvation, or logging issues specific to the Functions runtime.

Not applicable outside Azure Functions, and if you're on .NET's in-process model, note the hard November 10, 2026 support end date - this is a dated, real platform deprecation to plan migration around, not a hypothetical.

Inputs: the target language/runtime (.NET isolated, Node.js v4, Python v2), the trigger type (HTTP, queue, timer, Durable orchestration), and the hosting plan (Consumption, Premium).

Outputs: a working function using the correct programming model template, Durable Functions orchestrations for long-running or fan-out work, and fixes for the documented production issues - HTTP timeout workarounds, HttpClient reuse, async-throughout code, in-process-to-isolated migration, correct logging configuration, and cold-start mitigation.

Azure Functions (isolated worker .NET, Node.js v4, Python v2), Durable Functions, Application Insights, Azure Load Balancer (HTTP timeout behavior).

Developers building or troubleshooting Azure Functions who need the right programming model template plus concrete fixes for the platform's well-known timeout, cold-start, threading, and deprecation pitfalls - especially the looming November 2026 in-process model cutoff.