The hypothesis is that the OS doesn't allow opening a listening UDP socket multiple times on the same port. This could happen if the Receiver was stopped asynchronously and wouldn't shutdown and close its socket before a new test started.
The Travis build succeeded for the first trigger and then again for the second one, so by the principle of finite induction I infer it must work hereafter!
The hypothesis is that the OS doesn't allow opening a listening UDP socket multiple times on the same port. This could happen if the Receiver was stopped asynchronously and wouldn't shutdown and close its socket before a new test started.
Let's see if this change helps.