Error code 1 on calling `rclc_node_init_default`

aditya2592 commented 2 years ago

Issue template

Hardware description: Nucleo F746ZG RTOS: Zephyr 3.1.99, Zephyr SDK 0.14.0 Installation type: micro_ros_zephyr_module Version or commit hash: humble Transport: UDP IPv4 Agent: microros/micro-ros-agent:humble docker image

Steps to reproduce the issue

The issue is seen when the agent is already running on the host PC and the microROS device is started after. The microROS code first pings the agent to identify its presence and then proceeds to creating microROS entities. Could this in anyway be connected to HARD_LIVELINESS_CHECK? Do we need to have this check present to avoid the agent from sending invalid data to microROS and hence microROS failing to init nodes?

Expected behavior

No error codes on calling rclc_node_init_default

Actual behavior

Error code 1 on RCCHECK(rclc_node_init_default(&node, "zephyr_benchmark", "", &support)).

Additional information

Block of relevant code used

rmw_uros_set_custom_transport(MICRO_ROS_FRAMING_REQUIRED, static_cast<void*>(&default_params), zephyr_transport_open,
                                zephyr_transport_close, zephyr_transport_write, zephyr_transport_read);
while (true) {
    while (rmw_uros_ping_agent(ping_timeout, 2) != RMW_RET_OK) {
      LOG_ERR("Wait for agent...");
    }
    LOG_INF("Agent connected.");
    allocator = rcl_get_default_allocator();
    RCCHECK(rclc_support_init(&support, 0, nullptr, &allocator))
    RCCHECK(rclc_node_init_default(&node, "zephyr_benchmark", "", &support))

}

pablogs9 commented 2 years ago

Could you provide complete code for replicating?

I see a forever loop here, are you destroying the entities on the client side when the agent is no longer available? Check this approach.

pablogs9 commented 2 years ago

Any update on this?

nauticus-stu commented 1 year ago

I get the same behavior using the UDP transport. I just ran the example project in this repo targeting a board with an ethernet interface and making the appropriate changes to the zephyr_transport_params_t struct.

I was using Zephyr 3.2.0 and had to make some additional changes to the includes.

Debugging it I see UXR_STATUS_ERR_DDS_ERROR after the initial handshake with the agent in uxr_run_session_until_all_status at

uxr_run_session_until_all_status
run_xrce_session
create_node
rmw_create_node
rcl_node_init
rclc_node_init_with_options
rclc_node_init_with_options
rclc_node_init_default
main

aditya2592 commented 1 year ago

Sorry I forgot to update earlier. We are destroying entities and then init them again. The resolution we adopted for now is to just init in a loop until it succeeds. Usually it succeeds in 1-2 attempts.

nauticus-stu commented 1 year ago

I've narrowed the problem down to the agent. If I build the agent from source everything works as expected. The packaged agents don't seem to work. I'm guessing this issue can be closed, I don't think its a microros issue.

Hubery-wl commented 1 year ago

I have the same problem. May I ask how you solved it? Looking forward to your reply.

micro-ROS / micro_ros_zephyr_module