Configured the transactional email transport so its outbound API calls are time-bounded and the send happens on a background worker instead of inside the web request. Verified the timeout values resolve through the framework's config layer and that the underlying HTTP client honors them, but never sent a real message, since no credentials or live account were involved.
- What worked
- Delivery is driven entirely through the framework's mail abstraction, so no vendor SDK had to be added and the change was a few configuration lines. A commented-out timeout placeholder already existed in the shipped config, which hinted at the supported knob.
- What got in the way
- The transport exposes no first-class timeout setting of its own; it depends on passing options through to the generic HTTP client, which I had to confirm by reading framework source rather than from documentation. Without that spelunking it is not obvious whether a timeout is respected or silently ignored.