RTDE random deconnection

[dfl]

Hi,
I am using both UR-RTDE and RTDE C++ API to connect 2 or 3 clients (different process for testing purpose) on port 30004, to listen to robot variables (like joints, speed, tcp offset, …). And sometimes, I got an “End of file” ASIO error which suggests that the robot has closed all the sockets.

Does someone known what can cause this ? (Slow consumer, …)

For one of my listener, I just use the rtde_example ( Universal_Robots_Client_Library/examples/rtde_client.cpp at master · UniversalRobots/Universal_Robots_Client_Library · GitHub ) without sending anything to the robot (no send slider).

When I make a top command, the UR control is less than 90% and the bandwidth should be enough.
Regards,

[forbech]

It could be a problem with buffer overrun.

RTDE Connection Error [WinError 10054] Connection got closed from the Remotehost

Hallo, Setup: UR5e, RTDE Python Connection with Ethernet Cable and IP Port 300004 The robot and a laptop are connected with an Ethernet cable, and I tested my program which worked fine. After a bunch of test runs, it often started to close the connection as if the robot wasn’t connected anymore. Because there is a long process in Python, I increased the watchdog timeout in the UR Polyscope. That worked, but without changes, it started to give me the same error mid-running the program: [WinEr…

/Henning

[dfl]

Caused by a slow consumer ?

This is a track I will try to investigate this. My first investigation shows a low CPU usage on the PC who runs the test (less than 2% of CPU), and one client only receive and print joints, and timestamp. I do not see any delay (each frame are printed every 2ms).

[dfl]

Still investigate this one, using a simple C++ process only to receive data (DATA_PACKAGE), no processing.
Adding log, I see that must of the time I am waiting for data, and sometimes nothing come for a lot of cycle (125Hz), then suddenly I got a lot of data (filling up my socket buffer).
Properly process them (discarding the oldest and keep only the last one), then I can be disconnected due to buffer overrun (?).
I wonder if I can see a log in the robot itself when the data are sent exactly.

(more than :

2025-10-22 16:49:39.758 ERROR - RobotState messages are queued up: 7 {thread: RobotState - PostMan , loggerClass: com.ur.monitor.RobotState$PostMan}
2025-10-22 16:49:42.005 ERROR - RobotState messages are queued up: 3 {thread: RobotState - PostMan , loggerClass: com.ur.monitor.RobotState$PostMan} )

I will continue my investigation.