ASP.NET Core Middleware Bug Caused by Unsafe Response Stream Handling

A development team experienced repeated API outages traced to a middleware component that attempted to read the HTTP response body after it had already been sent to the client. The root cause involved response streams being consumed prematurely and accessed concurrently without thread safety, causing corrupted output especially with Gzip-compressed responses. The fix involved replacing the response stream with an in-memory buffer before pipeline execution, then safely reading, logging, and copying the response back to the original stream. Proper handling of content encoding was also added to correctly decompress Gzip responses before logging. The team noted that such bugs are rarely caught in local development and only surface under concurrent load in production environments.