We are interested in helping to speed up VDI logins with FSLogix but want to make sure we have a robust file services option. But the storage requirements worry us. What is everyone using to provide file shares. We are worried about Windows Servers and reboots for patching ripping out the user's VHD drive from machine.
Blogpost on requirements:
You're correct that rebooting the Windows Server that's hosting the fslogix (profiles or containers) will basically crash all VDI sessions. I don't use any clustering or HA for the file servers, so instead I've configured it so Windows Updates only run on weekends with no auto-reboot. On Horizon, I have all VDI set to auto log-off after about 18 hours, which allows the sessions to stay connected during the week but log out on Saturdays. Sometimes I will manually log out sessions during maintenance windows before rebooted the file server. It's not perfect, but I find it much better than dealing with many VDI's with massive OST files.
Since I only use Fslogix for Office containers and not profiles, the worst that can happen during a file server reboot is everyone loses their Outlook ost and Office cache files. It's not a big deal and sometimes I have to manually delete the vhdx files if they get corrupted anyway.
However, using Fslogix for profiles is a different beast. For profiles, I'm currently using OneDrive for Business with Files-on-Demand, along with Vmware DEM. This allows a user to login to a new profile, and items appear on the desktop as they get downloaded while DEM handles stuff like printers and personalization/settings. It's also not a flawless solution since OneDrive can be finnicky, but most users don't even notice a difference.
We are currently working on moving our users from LTSC Instant clones to SAC instant clones and the management desided to go to O365 client with on-prem Exchange and Sharepoint etc. With this we have decided to go with FSlogix instead of AppVolumes Writables. The deployment seemed to be pretty easy but we are running into couple of snags now as we are onboarding more and more people.
We have 2 Horizon PODS in 2 datacenters. We are using home sites for user assignments. We have deployed two file servers one in each site and I have split my GPOs to point clones to their local file share that hosts profiles. We are doing FSLogix profiles and also using cache mode for mailboxes. Also, we have done DFS replication between those two shares so in case on one site/pod down users can log in to another site and still get the same settings.
The problem started happening when we started onboarding more people as it seems that DFS is not keeping up, and I'm talking only 120 users. We didn't really do too many optimizations to fslogix and we are providing up to 20GB dynamically assigned as that includes mailboxes. Another problem is that if I all of the sudden get switched to another site, my OST gets corrupted, the same thing for OneNote. Deleting Outlook profile and recreating Notebook in OneNote fixes the problem but it is not ideal.
Also, I can definitely say that rebooting your File server midday will mess all of the clones that are using the VHD.
At this point we are looking into Clod Cache with FSlogix as this should help us disable DFS and allow client to synchronize it on both file shares at the log off. Also, it appears that it might be able to sustain file server reboot thanks to local cache but I did not figure it out yet
Definitely open to any other recommendations from fellow VMware Horizon admins
@LukaszDziwisz Active/Active DFS-R is not supported for profiles: https://docs.microsoft.com/en-us/archive/blogs/askds/microsofts-support-statement-around-replicated-.... The same is true for Folder Redirection shares.
We have been trying to do the same thing for a while now.
Two physical sites
Horizon with Win10 instant clones
and FSLogic configured for cloud cache.
However we have found that using App Volumes and configure FSLogix for Cloud Cache, our VDI sessions a. will not always logout and/or shutdown/reset and b. 99% of the time the users vhd.lock file remains. This prevents them from logging in again unless they wait > 5 minutes (sometimes they can't login at all until this lock file is deleted from the file servers).
Cloud Cache is pointing at one Windows file server (not using DFS) at each site.
If we not associate App Volumes with the user then everything works great, just no profile management.
If we deploy App Volumes with FSLogix writing to one Windows file server then everything is fine except we have no HA for the users profile which is a requirement for this project.
Running the latest version of everything that is available as of 3 weeks ago and have created a completely new deployment and same issues exist.
Have logged support tickets with both VMWare and Microsoft, which have now been closed with no resolution.
I would be very keen to hear how you are going and if you have found the same issue.
The other thing we have found is that if we associated App Volumes with an OU for computers, FSLogix with Cloud Cache results if login failure as it is unable to attach the users VHDX file.
Interesting reading the different approaches. I have been interested to compare UEM and FSLogix at some point. We use Ivanti for personalisation and desktop customisation. Plus a heavy thinapp user as well.