We have a two node Server 2019 cluster running the Scale Out File Server role for Storage Spaces direct. We use this to host our FSLogix Profiles for Remote Desktop Services. The cluster runs nothing else and is on the latest updates.
After about 30 users are signed into RDS the cluster CPU seems to spike the System service and will spike to 100% and sometimes stay up there for a while. We've disabled all Antivirus, uninstalled Defender firewall, updated all drivers and firmware, and rebooted multiple times. When the issue first started to occur we thought it might be our disk IO so we put in double the disk and went from 6Gb Sata intel DC SSD drives to 12Gb SAS SSD drives with way better performance. Issue still occurs. The strange thing is we had an error about the max envelope size being hit so we increased it in the registry, this fixed our issue for about 10 days, then suddenly it came back. This time we don't see any errors like we did before about the max envelope size being exceeded. The cpu in each server is a Intel E5-2670V3 2.3 ghz 12 core CPU. Since this is only a file server I would assume this would be plenty. Before this all started happening we would have about 85 users signed into RDS at once with 1-3% CPU usage on the cluster. Now with about 30 users the average CPU is around 30% with spikes to 100% and sometimes holding there.