I am trying to run 500 concurrent request in multithreaded and as well as multiprocessing mode by connecting CB N1QL Python SDK . while my process running I don’t see any activity in CB UI Console Dashboard about Disk Fetches Per Second I just want to know if CPU / IO or Memory is bottleneck . I can see CPU and memory trend in Server Statistics but Nothing is visible against Disk Queues . Is this a bug ? or am I missing something ? I am running CB 4 node all services in K8s cluster .
I believe that Disk Fetches are only necessary if the data being fetched is not already in memory. If all your data fits into memory (i.e. if “active docs resident %”, is 100%) then it would make sense that the number of disk fetches (assuming the cluster is in a steady state) would be very low.
@matthew.groves I have 200M docs spread across 4 data nodes … 37 GB allocated in RAM per node and most of them goes to store Metadata for keys and values . Also I have 550 GB in storage . I have Couchbase bucket and NOT memcache … So its highly unlikely that for a multiprocessing 3000 query scan I will not get any activity in Disk Fetches … Are you sure its not a bug . I just want to see some activity on Dashboard " Disk Fetches Per Second" chart