Unstable behavior of std::bad_alloc

[AssemSadek]

Bug report

Required Info:

• Operating System:
  • Ubuntu 20.04.5 LTS
  • Docker version 20.10.21
  • Jetson Xavier AGX with jetpack 5.1
• Installation type: docker image on arm64v8/ros:foxy-ros1-bridge-focal
• DDS implementation: Fast-RTPS

Steps to reproduce issue

1. docker run -it --rm --net host arm64v8/ros:foxy-ros1-bridge-focal bash
2. source ros_entrypoint.sh
3. ros2 run ros1_bridge dynamic_bridge --bridge-all-topics

Expected behavior

Actual behavior

terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc

Additional Information

This behavior is not reproduced at each time. But it happens frequently at unexpected moments. But I noticed some triggers:

1. at the moment of launching the node
2. at the moment of a new subscriber to a topic. The same topic can have already other subscribers or zero subscribers before this new subscription. Therefore, the number of subscribers is not an indicator.
3. whether I use the flag --bridge-all-topics or not, the behavior can pop out in both scenarios. Therefore, the number of bridges done is not also an indicator.

[F1iX]

For us, the std::bad_alloc problem appeared when using a ROS2 Foxy bridge to transfer sensor_msgs/NavSatFix messages from a ROS2 Iron container. Looks like the ROS2 versions are not necessarily compatible.

[gavanderhoorn]

Iron is not compatible with anything else IIRC. Mostly/largely because of https://github.com/ros2/rmw_dds_common/pull/68.