Closed fnunnari closed 3 years ago
Well. I guess this is not a VSM bug. I'm pretty sure the engine is thread safe. Either it is a java bug (we had this before from time to time) or it has something to do with ROS. I
ROS is not involved. Actually, the fast loops happened because the ROS machine was off. If it is related to Java, better be careful with the following version:
% java -version
openjdk version "13.0.4" 2020-07-14
OpenJDK Runtime Environment Zulu13.33+25-CA (build 13.0.4+8-MTS)
OpenJDK 64-Bit Server VM Zulu13.33+25-CA (build 13.0.4+8-MTS, mixed mode, sharing)
Investigated. It is a bug internal to ROS when executing the shutdown of the services.
I experienced an exception when two nodes, bouncing on each other with epsilon edges, were executing with no delay:
In general, a light-speed bouncing between two nodes is not a desired case, and has to be avoided. In fact this happened by mistake during development and testing.
However, I assume that the node execution should be thread safe also in this conditions. At worse, the system could slow down, but not crash.