VPC-RC taking upwards of 10 minutes, to hydrate its trunk cache when restarted at larger node size #634
Description
What would you like to be enhanced:
Why is the change needed and what use case will it solve:
Did experiment with 1700 nodes, restarted vpc-rc and track time it took to send event to that it manages this node, and calculated summary statistic for latency
Some raw numbers
=== Summary (seconds since T0 -> latest NodeTrunkInitiated per node) ===
nodes_total=1700 nodes_with_event=1700 nodes_missing=0
mean=384.334 p50=384.500 p90=667.200 p95=703.000 p99=732.000 max=737.000