Comment 3 for bug 1009793

The first test succeeds. The second test fails, when it attempts to bind "tcp://127.0.0.1:9502", with "ZMQError: Address already in use", which then gets wrapped into RPCException("Could not open socket").

So the obvious conclusion is that we're not closing ZMQ sockets after we're done with them.

If I comment out the first test, then the second test succeeds and the third test fails, supporting that conclusion.

But I don't see why the cleanup code fails yet. I see code that does attempt to clean them up (tearDown calls ZmqBaseReactor.close, which iterates over its sockets and calls close() on each.) And it's being called.