Another step on the road to dynamic scalability. The big feature is reallocation of traces to new Refineries during shutdown and after scaling events, which means that Refinery no longer corrupts telemetry under scaling.

There are also several other quality of life improvements, like Redis Cluster Support.

Also two new features:
SpanLimit for preventing large traces from overrunning memory
In and NotIn operators in rules