Fixed by https://github.com/sashafrey/topicmod/commit/0d7547d771c15558a40682b3a3fc4189473fd25b
The problem turned out to be simple. RPCZ uses function zmq_pool(), which accepts a timeout value. The documentation for zmq_pool() says 'timeout should be in microseconds', and this is what RPCZ does. In fact the zmq_pool() documentation is wrong, and the timeout shell be specified in milliseconds.
Fixed by https://github.com/sashafrey/topicmod/commit/0d7547d771c15558a40682b3a3fc4189473fd25b The problem turned out to be simple. RPCZ uses function zmq_pool(), which accepts a timeout value. The documentation for zmq_pool() says 'timeout should be in microseconds', and this is what RPCZ does. In fact the zmq_pool() documentation is wrong, and the timeout shell be specified in milliseconds.
The bug of zeromq documentation is discussed here: https://github.com/zeromq/jzmq/issues/75