-
Notifications
You must be signed in to change notification settings - Fork 81
Description
Generated by Generative AI
No response
Operating System:
Ubuntu 24.04.3 LTS with custom kernel 6.6.87
ROS version or commit hash:
jazzy
RMW implementation (if applicable):
rmw_zenoh
RMW Configuration (if applicable):
connect/endpoints={peer: ["tcp/some_other_container:7447"]};listen/endpoints={router: ["tcp/0.0.0.0:7447"], peer: ["tcp/0.0.0.0:0"]};scouting/delay=1000
Client library (if applicable):
rclcpp
'ros2 doctor --report' output
ros2 doctor --report
Sorry I can't provide this as it contains a lot of proprietary informationSteps to reproduce issue
Unfortunately, I don't have a good reproducer example. We were evaluating the feasibility of switching to rmw_zenoh for our production use-case, as it would greatly help us bridge connectivity across a mesh of containers compared to running the cyclonedds bridge.
When recording the /rosout topic, this message appeared after some time during operation:
[ERROR] [rmw_zenoh_cpp]: SubscriberCallback triggered over 10/rosout/rcl_interfaces::msg::dds_::Log_/RIHS01_e28ce254ca8abc06abf92773b74602cdbf116ed34fbaf294fb9f81da9f318eac.
And our recorder process actually failed to record most of the messages: only log messages from one specific node were received. Note that this recorder process is basically a ros2 bag record process, but customized to record presets of topics of interest for our use case.
Note that around the same time, we also saw some issues with services failing:
[ERROR] [rmw_zenoh_cpp]: Query queue depth of 10 reached, discarding oldest Query for service '/some_service'
Which suggests that something went wrong at the messaging layer beyond just /rosout.
Expected behavior
Our recorder process should record all messages flowing on /rosout, just like the cyclonedds case.
Actual behavior
No messages were recorded.
Additional information
No response