[docs] Tuning guide for the MDCR setting according to the network bandwidth between local/remote clusters
mocchira opened this issue · comments
As I commented on #1175 (comment), the default mdc_replication.size_of_stacked_objs can be too large for env with narrow bandwidth between local/remote clusters so that can result in filling the queue leo_sync_obj_with_dc_queue up and making LeoStorage under high load. I think we have to provide some tuning guide to set an appropriate number to mdc_replication.size_of_stacked_objs for those using MDCR.