These are chat archives for petabridge/akka-bootcamp

Nov 2016
Nov 07 2016 18:32
I'm working with deploying an actor in a clustered environment. I deploy an actor based on a message I receive, then immediately forward the message to it. I cannot deploy the actor ahead of time. Seems like the message goes to deadletters because the actor hasn't been fully deployed. Once the actor has been deployed, the messages flow to it. How do I overcome this?
Alex Michel
Nov 07 2016 18:32
I have latency spikes when consistent hash pool is rebalancing/resizing. Is there a way to reduce the chance of rebalancing and warmup it for peak load in advance? I tried to play with default resizer lower/upper and thresholds as well as give number of instances in constructor. Should I try to reduce the virtual nodes factor?
There is plenty of resources on machine so no problem to consume more cpu on average to be ready for peak