[Clarification Needed] Do in-flight HTTP requests and running function executions complete during slot swap?

[Tech-Fello]

Question

I'm seeking clarification on the behavior of in-flight function executions during a deployment slot swap for Azure Function Apps.

The current documentation states:
"Traffic redirection is seamless; no requests are dropped because of a swap. This seamless behavior occurs because the next function trigger is routed to the swapped slot." Then says right after "Currently executing functions are terminated during the swap."

Source: https://learn.microsoft.com/en-us/azure/azure-functions/functions-deployment-slots

These two statements appear contradictory:

1. "No requests are dropped" suggests in-flight requests complete
2. "Currently executing functions are terminated" suggests they are killed

Specific Questions

1. HTTP Triggers: If an HTTP-triggered function is mid-execution when a slot swap occurs, does the request complete on the original slot (since it has an established TCP connection), or is it terminated?

2. Timer/Queue/Service Bus Triggers: Do background triggers behave differently from HTTP triggers during a swap, since they don't have a client connection holding them open?

3. Clarification on "terminated": Does "currently executing functions are terminated" refer to:

   • All running functions regardless of trigger type?
   • Only background triggers (timer, queue, etc.) but not HTTP triggers with open connections?

Why This Matters

Understanding this behavior is critical for:

• Designing retry logic
• Determining if we need connection draining strategies
• Making architectural decisions about long-running HTTP operations