> OK until we implement the lifetime callback approach and works as expected.
>
> I think it's worth having an acceptance test for this now (i.e., that clients
> die when the server dies), since even with the lifetime callback approach I
> expect this to be the default behavior.
I don't really want to write an acceptance test for the SIGTERM behavior as I plan to replace it with something less ugly.
If I promise to to MP a test in the next iteration can we land this?
> OK until we implement the lifetime callback approach and works as expected.
>
> I think it's worth having an acceptance test for this now (i.e., that clients
> die when the server dies), since even with the lifetime callback approach I
> expect this to be the default behavior.
I don't really want to write an acceptance test for the SIGTERM behavior as I plan to replace it with something less ugly.
If I promise to to MP a test in the next iteration can we land this?