VMware View 4.6 Black Screen on Connect

[ETAVIS_GNS_AG]

Hi Everyone 🙂

One of our customer is using VMware View 4.6. We are now rolling out the thin clients and replace all the old worksations. some users are reporting "black screens" on connections. we are using PCoIP. but already checked the display driver version, and all other KB articles about that.

the configuration of the vmware view client is done by active directory gpo's.

what i see on the event of the connection server is the follwing on each connect.

first i see de:User domain\user requested desktop null

and after that the right desktop pool name. i havent found anything about that, any ideas?

for the black screen I also created a ressource pool with a small CPU reservation for the desktops. so they will not fall to "sleep" and need to long to get cpu time.

[griffinboy]

Have you made sure the 3 essential points addressed on this page are satisfied?

I had the black screen issue too (even after following the instructions on that page to the letter) and it ended up being my load balancer configuration. I was missing a virtual server for TCP 4172 (PCoIP). I only had the UDP one setup. After creating it and making sure the points on that page above were all OK it all started working like a charm.

I hope this helps.

[ETAVIS_GNS_AG]

Hi griffinboy

Thanks for your answer. The connection is just on the internal lan. We are using two connection servers (standard & replica). Connecting on these is made without a NLB right now. Just connecting on a DNS record with two IP’s of both of them. So there is no problem to establish the TCP or UDP connection for PCoIP.

The connection runs mostly of time like a charm, but just sometimes the black screen occurs. Once at 40 logins…

Regards angelo

[griffinboy]

Oh I see. Do you know if you are hitting the same connection server   every time it fails? Or does it happen randomly when connecting to   either of them?

By what you were saying in your first post (the entries in your event DB) it could be that the problem might exist on one of your nodes only. I guess  the easy test for that is to disable one of your connection server,  connect to the other one over and over again, and see if you see the  error again eventually. If you do then try the same thing on the other  one. This way you can tell which node is the culprit. Maybe you have tried that already?

If the problem happens on both then it could be a long debugging process ahead of you...

[bjm534]

Check to make sure that your PCoIP gateway settings are configured correctly. If you go into the VIEW admin console, select servers on the left side, and highlight a connection broker and click edit. You will see two fields, one that has the VIEW URL and the other is the PCoIP gateway address. It is very important that these are set correctly when doing load balancing. If your URL internally is VIEW.domain.com then both of your connection servers should have that as their URL. As for the PCoIP gateway if you are tunneling your connections through the connection brokers with the SSL gateway functionality then you'll need to put in the Virtual IP given to your NLB cluster as the IP, otherwise you'll have to use the individual IP's of the NLB servers. You may have to set the PCoIP value this way enven if you aren't tunneling it depends on the way that the NLB is setup.

-Brad

[ETAVIS_GNS_AG]

Hi

We are not using NLB, just two connection servers. We are using DNS round robin right now.

hostname1 -> points to ip of connection server 1

hostname2 -> points to ip of connection server 2

view.domain.local -> resolves to both ip's of the connection servers.

the vmware view client's connects to view.domain.local. so i have to setup the pcoip gateway settings to this hostname?

[griffinboy]

On the page I linked to in my previous post, there's a video at the bottom explaining exactly how it works and should be setup. The first part deals with remote connectivity over PCoIP which is of no interest to you, but the last part shows how to do it internally, without Security Servers, and in a resilient manner.

Correct me if I'm wrong but I would have thought session persistence is pretty important in a VDI environment and therefore simple DNS round Robin may not be sufficient, hence the recommendation about using NLBs.

[markbenson]

ETAVIS_GNS_AG wrote:

Hi

We are not using NLB, just two connection servers. We are using DNS round robin right now.

hostname1 -> points to ip of connection server 1

hostname2 -> points to ip of connection server 2

view.domain.local -> resolves to both ip's of the connection servers.

the vmware view client's connects to view.domain.local. so i have to setup the pcoip gateway settings to this hostname?

Following http://communities.vmware.com/docs/DOC-14974 will solve this.

Mark

[caryers]

Do you have any legal disclaimer in your AD login process? I ran into this same issue in our environment. First, I had to make sure all the ports were added to the rule to allow VDI access. Those ports for our environment are:

80, 443, 4172, 32111,1688, 8009, & 3389.

Since you have to select OK on our legal disclaimer, I had to use a new GPO that did NOT have that legal disclaimer within AD. I then placed our county required legal disclaimer within the View global settings area.

That resolved our black screen issues.

Scott

[ETAVIS_GNS_AG]

Hi all

Thanks for your replies. With your posts, a connections should never be possible. but we have two blackscreens on about 100 working connections.

so it is not a general config issue or something like that.

has anybody experince with the new vmware tools, which brings a new display driver version x.x.51 wit it?