Israel Lot

In what setup are you running? The fork operation failed. In the windows port, as there's not a kernel fork like Linux, it means a copy-on-write operation involving the pagefile...

A small dataset makes less likely the pagefile theory. Please share the whole config file used. Are you using AOF or RDB or both?

@neuecc any thoughts on this? Let me know if you would prefer a direction other then simply adding a two parameter callback option. I believe another feasible solution would be...

Maybe try using **PostPublish** instead of **PublishAsync** I guess that would align better with the behavior you had in mind. **PublishAsync** will await on network, **PostPublish** is fire and forget.

I've ran your example @80dB. With this loop I get around 30k/s as you mentioned ``` while (!cancellationToken.IsCancellationRequested) await conn.PublishAsync(key, "HELLO WORLD"); ``` With this one I get >1M/s ```...

> @israellot as I mentioned, PostPublish does get me equal performance (1M/s is impressive, I was getting around 720k-750k/s running on Windows with a 12900k CPU), but have you checked...

It's also a lot cheaper to use JetStream vs a managed queue like ServiceBus.

Should be open again

The problem is the format specified in the sprintf. Our printf doesn't process the %.0f. The solution is quite simple.

I've just pushed a commit that should fix it. Thank you.