Currently, I have thousands of workers starting simultaneously and performing tasks, which creates pressure on the database. After reading the documentation, I discovered that it’s possible to limit the QPS (Queries Per Second) for the cluster’s service. What effect does changing this QPS limit have on the service’s internal scanner, and are there any other impacts
Can you give more info on the problem? Can you show persistence latencies by operation:
histogram_quantile(0.95, sum(rate(persistence_latency_bucket{}[1m])) by (operation, le))
and resource exhausted:
sum(rate(service_errors_resource_exhausted{}[1m])) by (operation, resource_exhausted_cause)