whibr's Posts

I am using Horizon 7.8 instant-clone, Windows 10 1803, with App Volumes 2.16.  Horizon client testing with is 5.0.0 for Mac. When a user connects to a desktop session using Blast protocol and the Horizon Client through a UAG 3.4 gateway, the login display will "hang" at the Welcome message for up to 1 minute or so before the Windows desktop appears. If the user logs in from from the same client but selects PCoIP as the display protocol, the desktop session login proceeds quickly, Welcome animation will stop for about 2 seconds then proceed, and the Windows desktop appears after about 18 seconds.  So it seems to appear Blast and UAG are not working together correctly. Has anyone seen this before? Connecting directly without a UAG in between takes about the same 16-18 seconds for either Blast or PCoIP protocols.

I am provisioning Autocad Civil 3D 2019 right now, and it has just past the 2 hour mark ... The task manager CPU utilization is almost flat-line around the 0 - 1% range, but I do see a small amount of activity with the appstack disk.  See screenshot. It isn't just Civil 3D that I am seeing this slowness. I'm having to deal with it on many of our engineering applications that I am attempting to provision. Some provision fairly quickly, however. I'm using: Windows 10 Enterprise 1803 with current Microsoft updates. disabled or stopped most unnecessary services, such as Defender. App Volumes 2.16 vSphere ESXi 6.7.0 build 13473784 vCenter Server Appliance 6.7 U2 (6.7.0.30000) All-flash vSAN, NVMe-based Dell R740 server cluster Installing the app the same provisioning VM without App Volumes provisioning mode doesn't take very long at all. Something is severely killing provisioning and I'm not sure where to look. I must be missing something obvious. Suggestions? Thank you. Brian

When I minimize a Horizon client 4.8 BLAST session window, after just a few minutes my instant-clone desktop (Windows 10 with Horizon agent 7.5.1) will disconnect my session.  With a PCoIP session, the session will remain connected after minimizing the session window.  Is there an idle session timeout setting for BLAST, or is there something else I should check?  I don't want a user's BLAST session to automatically disconnect if the user has paused work for a period of time.  Thanks.

The Horizon agent should not be installed on the system being used to provision an appstack, correct?  What will happen if I leave it installed?  So I had provisioned an application (Bluebeam Revu 2017) successfully without the Horizon agent installed as recommended. When I then attached and use the application in a Horizon desktop pool with NVIDIA GRID vGPU enabled, the application would have graphics issues and some of the text box and sticky note features wouldn't work.  The application did work just fine when I tested it on another pool that didn't use vGPU, but I need the appstack for our vGPU-enabled Horizon desktop pools. I next tried provisioning this app on a vGPU-enabled desktop pool parent VM and keeping the Horizon Client and direct-connect plugin in the VM.  This is needed to activate the vGPU graphics in-session.  This approach did fix something, since the application is working correctly when I attach it to a vGPU pool user session.  During installation when provisioning, I think the Bluebeam installer is doing some type of optimization and profiling steps with the graphics capabilities.   I'm just wondering if anyone else has experience provisioning appstacks for attaching on NVIDIA vGPU desktop pools, and if there are any pitfalls I should be on the lookout for by doing it this way (ie. keeping Horizon agent installed so I could activated the NVIDIA graphics).  Thanks.

I have kerberos authentication method added and working to our default_access_policy_set.  We have a multi-domain Active Directory environment, so when internal users go to the workspace FQDN, they must select a domain from the drop-down menu and click the Next button, and then kerberos logs them in automatically.  If the user mistakenly selects the wrong account domain, the next page shows "Access Denied. You do not have access to this service..." and there is no way to go back and select the correct domain.  I am able to easily clear the browser cache to allow the user to select a different domain, but why can't we just use UPN for login and skip domain selection drop-down?  Has anyone been able to get around the select domain requirement for user login? I would like to get rid of the domain selection process and just allow the user to enter their userPrincipalName to log in.  The system should automatically be able to authenticate their domain credentials from just the UPN.  I haven't been able to get that to work with AD authentication and kerberos single-sign on, though.

I am doing basic planning for upgrading our Horizon 6.2.1 environment to Horizon View 7 soon after GA.  The compatibility matrix shows the following for ESXi and Horizon View 7.  Only ESXi 6.0 U2 is supported with Horizon View 7? ESXi and Horizon Matrix: I guess this means that I shouldn't risk upgrading our ESXi systems just yet. Apparently, just vCenter Server 6.0 U2 is supported with Horizon View 6.2.1, so I will need to wait until Horizon View 7 GA and then plan a longer upgrade window to bring both Horizon View and ESXi current at the same time...or am I just missing something? Maybe the matrix will change upon Horizon 7 GA. vCenter Server and Horizon Matrix:

I am wanting to create App Stack's containing graphical CAD design applications that we will assign to users logging in to a vGPU-enabled non-persistent desktop pools. Has anyone provisioned such applications in an AV version 2.9 App Stack?  I have been successfully provisioning and deploying apps like MS Project/Visio on VM's that aren't using vGPU desktops for some time now.  With App Volumes, though, I think I remember reading in the past that you should not install the Horizon agent on the provisioning VM,  If that is still the recommendation, then I won't be able to activate the vGPU environment on the provisioning VM when installing my applications since that requires a Horizon PCoIP session (and NVIDIA display driver in the guest).  Maybe the applications would still work just fine if I use my other soft 3d provisioning VM and assign the stack to user on a vGPU vm, and I can test that, but I wanted to first ask if anyone else has been down this road yet and could share recommendations.  Thanks.