Lifecycles hangs on activate, in docker container
Hi ROS community.
I have the problem where my lifecycle nodes will hang on activation, but only if i use more then 2 nodes. The nodes does not hang every time, they seem to hang at random, though with the example docker below it is repeatable. Its likewise annoying as launch a new node, in its own container, seems to activate the previously stuck nodes and allows for them to finish their configuration.
In order to demonstrate the supposed bug, i have created a minimal reproducible example: example
I'm unsure how i would proceed with debugging this, as there is no prints to the console. For versions, the docker is build with Ubuntu 22.04 and the ROS2 version is humble. FastDDS is likewise the newest version.
I have however reproduced this with the binaries from the official repository, and the bug is especially present on low power systems such as a Jetson nano.
Is this a known bug? Or is there a know workaround?
Asked by HansLarsen on 2023-07-14 20:12:07 UTC
Comments
Try cyclone. This is a relatively known issue in Fast DDS that persists today.
Asked by stevemacenski on 2023-07-17 01:21:18 UTC
Unfortunately the platform i'm debugging for, already has users, who is using ROS2 and fastrtps, so i was hoping to debug it and potentially create a fix of some sort. Do you know if there are any github issues with this problem?
Asked by HansLarsen on 2023-07-18 17:22:43 UTC