The error you described happens when the client exhausts it’s internal pool of TCP connections. This is usually an indication of either high-latency, very high OPS, and/or perhaps very high concurrency - all three can be related. The solution is to increase the pool size (MaxSize) until the error goes away, exactly as you did…it’s just a matter of tuning the client to your environment.
-Jeff