Horizon 6.1 environment with Vsphere 6 cluster, Win7 64 desktops keep crashing/rebooting

[TexRob]

I manage about 7 different Horizon View environments for customers, some on prem, and some in our hosted environment.  I have had no problems (lately) with any of them, and some of them exist on this same Vsphere environment, and one other is on the same Horizon View release.  The only discernible difference is I believe the environment not having issues is on hardware version 7, and the problem environment is 11.  The issues are happening across all servers in the cluster, and the cluster includes other working View environments, using the same storage array in some circumstances.

So, the issue is that these Win 7 64 desktops are crashing or rebooting, and I can't figure out from the log exactly which.  I don't see "WinBSOD" in the log.  The users report the issue after stepping away for a few minutes, and they come back to a black screen, move their mouse to try and wake it, and usually lose connection.  It looks like maybe there is a video issue, but I can't really figure out what or why.  They all have 35mb of video memory, and most users are running 1920x1080 or lower on a single monitor, so this should be more than enough.

Attached is a log from one of the machines having the issue.  Anyone have any insight?  able to point me in the right direction?  I usually figure this stuff out on my own, but this one has me pretty stumped.

Thanks

[vishalchand]

We have similar issue for some users have reported VDI desktops black-screening and disconnecting with View 6.2. VMware has introduced a new video driver (version 6.23) in View 6.2 that greatly improves speed and quality, however to utilize this fully we need to be on vSphere 6.0 Update 1a. The other fix would be replace the video driver inside the guest OS of the gold image/persistent desktops which is a difficult process

[h3nkY]

How many MB your VRAM size?

Increasing VRAM size may solve the issue.

[nonsparker · Accepted solution]

We are having the same issue, although it's not widespread. VMware Support pointed us to this KB http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=211724...

I have yet to make the change but I thought it might help you.

[TexRob]

Thank you everyone for the responses.  I'm going to start with the workaround posted on that one KB and go from there.  I will come back and post my results and mark the correct answer/s.

Thanks,

[jmacdaddy]

I was experiencing this issue with one of my clients and only where the thin clients were connected to high res monitors.  The workaround suggested in this KB did not work and VMWare support confirmed that the workaround doesn't work consistently.  For me the fix was ESXi 6.0 Update 1 and View 6.2

[TexRob]

I was experiencing this issue with one of my clients and only where the thin clients were connected to high res monitors.  The workaround suggested in this KB did not work and VMWare support confirmed that the workaround doesn't work consistently.  For me the fix was ESXi 6.0 Update 1 and View 6.2

Ugh.  So, I have applied this workaround to two users, but the value it says to change to false was not even in their VMX files.  I added it, but since the KB does not reference what to do in the case of it not being there, which they would normally do, it makes me less than confident this will work.

The problem with upgrading is I have 5 customers on various versions below Horizon 6.2, so I am going to have to update them all first, then all my hosts, just to fix this issue on their one environment that is only affecting a few users.  Good times.

I'll post back once I see if the workaround works for them.  Marked correct answer because that KB is definitely my issue, the log matched exactly.

[TexRob]

FYI, the workaround worked for me.  Recapping because it's notable, that entry was not in the VMX file, and adding it seems to have fixed the issue for my users.  Long term I will be updating all my software to the latest Horizon and ESXi/Vcenter versions, but this seems to be working for now.

[jmacdaddy]

Did you apply the VMX fix on the template VM and then recompose or did you apply the fix directly to the VMX files of each linked clone?  I seem to remember an issue where the values would not stick on the linked clones after a recompose...