error message: "error sending response: Host unreachable" even if the service call behaved as exepected

[paoloelle]

Environment

• OS Version: Ubuntu 22.04
• Source or binary build? Binary Gazebo Fortress

Description

• Expected behavior: Some message that tells me if the service call behaved as expected even if an error occurred in sending the response to the service
• Actual behavior:
  I'm just implementing a simple scenario where I have a robot and a couple of simple objects that need to be moved in the arena with a certain frequency. I have a bash script that allows me to do that. The problem that I'm facing is that, even if the robot and the objects are moved as expected, I keep seeing the message [ruby $(which ign) gazebo-1] NodeShared::RecvSrvRequest() error sending response: Host unreachable when the timeout for the service is triggered (Service call timed out).

I guess the issue can be related to #243.

Steps to reproduce

1. Run a simple simulation with some objects
2. Change the position of the objects continuously

Output

This is the output that I get when the timeout for the service call expires (even if the object is moved to the new position as expected):

• [ruby $(which ign) gazebo-1] NodeShared::RecvSrvRequest() error sending response: Host unreachable
• Service call timed out

[azeey]

@caguero would you be able to take a look?

[paoloelle]

I’m trying to run the same example with more output information for debugging. Could there be a maximum number of ign service call commands that can be executed in a single simulation run? I tried the same also with Gazebo Harmonic and ROS 2 Jazzy but the problem seems exactly the same.

[paoloelle]

Follow-up of this

I’m trying to run the same example with more output information for debugging. Could there be a maximum number of ign service call commands that can be executed in a single simulation run? I tried the same also with Gazebo Harmonic and ROS 2 Jazzy but the problem seems exactly the same.

These are the results of the experiment mentioned above: it consists of stopping the simulation, moving the TurtleBot4 robot, moving the objects, and letting the simulation run for 10 seconds. The experiment is shown in the following video.

simple_sim-2024-12-20_09.56.47.mp4

I executed this simulation for ~24 hours, and these are the results that I got. In the following plots please don't take into account the 25th hour.

As can be noticed from the plots there is both a general decrease in the number of services called per hour and the successful rate per hour.

Please let me know if I have to open another issue related to this or we can keep this one