CI failing due to missing controller_manager

[fmauch]

trafficstars

Affected ROS2 Driver version(s)

Basically all PRs

Used ROS distribution.

Iron, Rolling

Which combination of platform is the ROS driver running on.

Docker

How is the UR ROS2 Driver installed.

Build the driver from source and using the UR Client Library from binary

Which robot platform is the driver connected to.

URSim in docker

Robot SW / URSim version(s)

The one from CI

How is the ROS driver used.

Headless without using the teach pendant

Issue details

Summary

Currently, some of our CI jobs regularly fail due to a missing connnection to /controller_manager

Issue details

It seems that sometimes our controller spawners seem not to be able to contact the controller_manager. For single controllers we get

 [spawner-8] [INFO] [1698650407.667882462] [spawner_scaled_joint_trajectory_controller]: Waiting for '/controller_manager' node to exist
    [spawner-8] [INFO] [1698650409.677460494] [spawner_scaled_joint_trajectory_controller]: Waiting for '/controller_manager' node to exist
    [spawner-8] [INFO] [1698650411.687458830] [spawner_scaled_joint_trajectory_controller]: Waiting for '/controller_manager' node to exist
    [spawner-8] [INFO] [1698650413.697357365] [spawner_scaled_joint_trajectory_controller]: Waiting for '/controller_manager' node to exist
    [spawner-8] [INFO] [1698650415.708340132] [spawner_scaled_joint_trajectory_controller]: Waiting for '/controller_manager' node to exist

And the tests finally fail because the controller never comes up.

Steps to Reproduce

Not sure how to reproduce this, as this happens in some pipeline runs, but not all.

Expected Behavior

It should work

Actual Behavior

e.g. https://github.com/UniversalRobots/Universal_Robots_ROS2_Driver/actions/runs/6689460108/job/18173039411

Workaround Suggestion

none..

Relevant log output

No response

Accept Public visibility

• [X] I agree to make this context public

[KKSTB]

Controller spawner has a wait for service timeout of 10sec only. Could this be caused by a newly started ursim where the boot time takes more than 10sec?

[fmauch]

Controller spawner has a wait for service timeout of 10sec only. Could this be caused by a newly started ursim where the boot time takes more than 10sec?

No, we specifically pass a timeout of 120 seconds in there. But thank you for the input @KKSTB.

[urfeex]

This got fixed upstream