We have a 2 node Windows 2012 RDS farm with UPD enabled. All UPD are on a dedicated Win2012 file server. Users documents, desktops, and appdatas are excluded from the UPD but are redirect to a different share on the same file server. Now and then, users complain of long logon times, often where the screen goes black for several minutes. When this occurs users which are already logged onto the RDS servers also report general performance issues issues and error relating to 'unable to access temporary environment, etc. To correct the issue we ask users to log out so we can reboot both RDS servers and the file server. Obviously this is far from ideal and causes a unplanned disruption/downtime. Once the servers have restarted users are then able to log back on to the RDS farm and performance is back to normal. The issue then reoccurs within a matter of weeks or a month. We've ran some perfmon logging on the three servers but couldn't see any obvious signs of resource contention. Strangely, if we log onto the RDS servers as an administrator, who are excluded for folder redir and do not have UPD, we do not witness these problems. We believe the issues is related to the UPD as we also have a legacy four node Win2008 RDS farm. Users on these servers do not have UPD and do not suffer from these issues despite having roaming profiles and redirected folders pointing to the same file server on a different share.
Has anyone come across a similar situation or have any suggestions on how to troubleshoot this?