GitLab

The code that waits for the daemon to have opened its port is essentially:

while True:
    try:
        s = socket.create_connection(addr, timeout=timeout)
        s.close()
        return
    except socket.timeout:
        logging.error(f"daemon did not respond at port {port} within {timeout} seconds")
        raise
    except socket.error as e:
        logging.info(f"could not connect to daemon at {port}: {e}")
        pass

Would it not make sense to have a time.sleep(0.1) instead of pass on the last line? It shouldn't affect the total time to wait in any significant manner, but would reduce the number of times the loop is iterated significantly.