Problem/Bug with reestablishing connection when restarting the publisher PC using ROS2 and FastRTPS.
We have a problem when running two ROS2 nodes on two different PCs using FastRTPS. The problem can be reproduced with a minimal example, with one of the ROS nodes simply publishing to a topic with default QoS and the other subscribing to that topic with the same QoS.
With both running, if we break the ethernet connection and reestablish the connection, the subscription is reestablished automatically as it should and the published messages are received shortly after.
If we shut down the either the subscriber or the publisher node, everything is reestablished perfectly as well.
If we restart the PC with the subscriber node, everything also works as is should.
But, if we restart the publisher PC, then the connection is only once in a while reestablished but most often not. If we restart the subscriber afterwards, everything works well again. Could there be something we are missing? We are wondering what the difference is between restarting only the publisher node and the whole PC, and suspecting it maybe to be caused by by a new instantiation of a multicast connection?
The problem is a bit critical to us, as it affects the our customers experience with our product.
