Azure Functions Flex Consumption Node 22: intermittent response-header stalls, including minimal handlers

Matt Wallace 0 Reputation points
2026-11-22T22:19:14.9366667+00:00

We are investigating intermittent response stalls in a non-production Azure Function App.

Configuration:

  • Flex Consumption, UK South
  • Node.js 22
  • 512 MB, maximumInstanceCount 1
  • No always-ready instances
  • @azure/functions 4.17.0

Symptoms:

TCP/TLS typically completes within approximately 0.10–0.16 seconds, but some requests receive no response headers before our fixed 30-second client timeout. Other requests complete in approximately 0.15–0.50 seconds. Curl reports exit code 28 for the timed-out requests.

Isolation performed:

  1. A minimal handler removed business logic, Graph calls, managed-identity calls and the MCP SDK. It returned only small static responses.
  2. Both buffered and streamed responses were tested. Of ten requests, one buffered and one streamed request timed out; the other eight completed.
  3. A later scheduled comparison using curl from both a Mac and Azure Cloud Shell passed all four requests from each client.
  4. A subsequent full-runtime readiness test, using curl HTTP/1.1 and a 60-second settling period, again timed out. Public unauthenticated requests and absent routes also experienced stalls, so downstream data access cannot explain every failure.

Portal diagnostics for 22 November 2026 show:

  • Seven Microsoft.AspNetCore.Connections.ConnectionAbortedException occurrences: “Exception encountered while listening to EventStream. The HTTP/2 connection faulted.” Latest timestamp 21:49:13 UTC.
  • Three System.InvalidOperationException occurrences involving WorkerTerminate: “Can't write the message because the request is complete.” Latest timestamp 21:14:48 UTC.
  • Unhealthy script-host/readiness entries reporting “No script host available”, errorCode NoScriptHost, at 20:49:37 UTC.

We deliberately deployed, restored and disabled the app during these experiments. Some runtime errors may therefore relate to shutdown or deployment rather than the request stalls. We have not established that they coincide.

The app is currently disabled. We have not established memory exhaustion, a platform defect or a need for increased capacity.

Questions:

  • What targeted diagnostics can distinguish host/worker communication failures from deployment/shutdown effects in this situation?
  • Are there known issues matching these symptoms on Flex Consumption with Node 22?
  • What is the smallest useful evidence collection before considering a hosting or capacity change?

We can provide resource-specific details privately through an appropriate Microsoft support channel.

Azure Functions
Azure Functions

An Azure service that provides an event-driven serverless compute platform.

0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.