Azure Functions insights

Application Insights Sampling Hiding Critical Failures

Default Application Insights sampling (max 20 items/second) can cause missed exceptions and request failures, obscuring critical production issues during troubleshooting.

External service failures (databases, APIs, storage) propagate to function failures without proper retry logic or circuit breakers, causing cascading failures and poor user experience.

Without telemetryMode set to OpenTelemetry in host.json, the Azure Functions Host and isolated worker process don't share telemetry context, resulting in broken distributed traces and missing parent spans for function invocations.

Azure Functions can fail to load after deployment with no visible errors in the portal, requiring log-based troubleshooting to identify configuration, dependency, or code issues that prevent function initialization.

Missing or misconfigured AzureWebJobsStorage setting, deleted storage accounts, rotated keys, or firewall restrictions prevent function host initialization and runtime operations.

Inefficient code patterns, long-running functions, or SNAT port exhaustion cause high CPU or memory usage, leading to performance degradation and function failures especially on Consumption plans.

Dedicated (App Service) plan functions without AlwaysOn enabled experience cold starts and increased latency similar to Consumption plans, negating the benefits of dedicated hosting.

Missing invocation history in Azure portal and Application Insights makes it difficult to identify when functions stopped triggering or failed silently, delaying incident detection.

Azure Functions on Consumption plans experience increased latency when idle functions restart, with default 5-minute timeouts exacerbating the issue for long-running initialization.