Are you having garbage collection stalls? I've seen such messages when a long GC stall causes some requests to time out the instant the stall is over, even though the reply was sent from the PLC (still in OS buffers, presumably).
No, you need to enable GC logging so you can see how long your pause-the-world stalls are taking. Tens of milliseconds is fine. Maybe an occassional 100ms is fine. Any more than that and short request timeouts can start falling within the stall, and thereby expiring.
Ok @Kevin.Herron , thanks. Running pinginfoview with a shorter ping period a found that you are right: there are some connection problems with the PLC. I'll ask the network admins.