Improbable Icon

Is there any solution for the lag on killing a worker?

worker

#1

From the Wizard tutorial:

SpatialOS stops the worker immediately. Just a few seconds later, the load-balancing algorithm kicks in, and it starts a new worker. Note that nothing at all happens in the game window - the client isn’t even disconnected. Load-balancing is completely invisible to players!

I’ve tried this. What happens to the client when killing a worker is a bit a lag, then “fast forward”:

Could not record a longer spike, but there happened a 2-3 seconds long one as well. This seems to be similar to the lag what we often see while playing Worlds Adrift.

Is there a solution to this? Like overlapping regions with redundancy, so losing a worker does not cause a visible lag?


#3

Would be interested in knowing this as well. But from everything I have seen and worked with its just something that has to be lived with / accounted for.

The simulation will need time to spin up a new worker and load everything up on it and / or to reconfigure all the entities onto other works.