I notice that there are some nodes (e.g. GroupBy and Joiner) that are capable of using multiple processors and much of the resources of a KNIME session. When run with big data, they also seem to occasionally lock the GUI when they go through certain parallel parts of their computations.
This is fine, until two or more of these nodes run at the same time. They often crash my KNIME session, though they can run fine independently
There are tricks (connect flow variable ports) and a setting to restrict to 1 thread. These have obvious disadvantages when I am using so many of these types of nodes, either having to connect them across a sprawling workflow or losing parallel processing capability.
Any suggestions would be welcomed.