All of our nodes are ingest nodes because that's the default 
However, only our hot nodes are targeted by most clients for indexing and only a subset of those would use ingest pipelines.
The stats appear to show that the node targeted by a client runs the ingest pipelines, does that sound correct?
If we wanted to add some "dedicated" ingest nodes and only target those for a specific high ingest rate pipeline app, would those nodes run the pipeline code?
I assume the heavy indexing cycles would still occur on the hot nodes?
Thanks