Browser downloads fill up write-cache

[Björn Schläfli]

Virtual Apps

PVS

50 GB Write-Cache Disk for every VDA multi-session

 

Some of our users download very big files (business need) in Edge browser. As they download it in their %temp% write-cache is filling up quickly. We've configured a gpo setting lately to recommend the download path to point to their user home share. Only recommended because users have to be able to change the download location. Some users changing the path to their temp again... 

How do you guys configure download paths for browsers? It's a real "danger" for write-cache, if big downloads are written to the user profile.

[Simon Graham]

You should use your profile management solution to manage user specific data like this. The cache disk really should not be used for anything apart from the cache and pagefile and we do not support using it for anything else.

[Björn Schläfli]

I use Citrix UPM. What do you mean Simon? While the user is logged in everything written to it's profile is written to C drive and if it's a lot of data the write-cache in ram isn't enough and it will use the overflow (e.g. a user maybe uses an application which writes data to user's %temp%). So it's nothing else than cache.

[Jeff Riechers]

Use the Exclusion list options  to exclude those temp files out of the profile.

 

https://docs.citrix.com/en-us/profile-management/current-release/configure/include-and-exclude-items/overview

 

Setup UPM to delete any local cached profiles at logoff.

 

Also with the larger delta changes with non-persistent images, I have taken to making the writecache drive match the c:\ drive 100%

[Björn Schläfli]

I think you don't understand me correctly. 

I have upm exclusions in use to exclude everything not needed (a user profile is approx. 150 mb in size). Also profiles will be deleted at logoff of course.

example:

user logs in, uses edge browser and has to download some vlc files needed for their work. These files have a size of up to 12 gb. If they choose %temp% as download path (although we set the path to user home drive as recommended setting) these files are downloaded to their local profile copy on the server and of course the write cache is then filled. When the user logs out the local profile is deleted and user's %temp% will not be written to profile share because it's excluded. 

 

The problem here occurs while the user is actively working on the server and writing files to his profile. I cannot remove write access to his profile ? and I cannot set the download path as a must because some of theme need to be able to choose where they want to save downloads.

 

Is this more understandable?

[CarlStalhood]

Have you tried configuring Folder Redirection for the Downloads folder and point it to a file share?

 

Or if you use a Profile Container solution (e.g., FSLogix), then Downloads will go inside the container that's on the file share instead of in your cache disk.

[Jeff Riechers]

5 hours ago, Björn Schläfli said:

I think you don't understand me correctly. 

I have upm exclusions in use to exclude everything not needed (a user profile is approx. 150 mb in size). Also profiles will be deleted at logoff of course.

example:

user logs in, uses edge browser and has to download some vlc files needed for their work. These files have a size of up to 12 gb. If they choose %temp% as download path (although we set the path to user home drive as recommended setting) these files are downloaded to their local profile copy on the server and of course the write cache is then filled. When the user logs out the local profile is deleted and user's %temp% will not be written to profile share because it's excluded. 

 

The problem here occurs while the user is actively working on the server and writing files to his profile. I cannot remove write access to his profile ? and I cannot set the download path as a must because some of theme need to be able to choose where they want to save downloads.

 

Is this more understandable?

We got you.  So if users are breaking out of locations you have selected for things, you may just have to boost the size of the writecache drive. I had to do that recently for a customer that worked with monstrous data sets.  No amount of redirection worked.  Bumping up the writecache drive and thin provisioned in the hypervisor was the only fix.