timeout gets doubled in 3.1

[Carlos Carvalho]

The timeout value actually gets doubled in 3.1 (didn't check in 3.0).
It first happens between sender and destination, producing an error
message. However in the destination one of the processes continues to
run, and only times out and exits after another timeout interval. So
in practice all processes terminate only after twice the period.

Is this intentional? Is it unavoidable?