Why CloudXR connect fail because Adapter 0 not found in the list of heads?

Gaming and Visualization Technologies XR – VR/AR/MR CloudXR (VR and AR Streaming)
1

hello everyone

When I test CloudXR, the connection always fails with GPU T4，
I compared my computer（GPU RTX 3060） with the problem environment（GPU T4）. I found the problem environment has the follow error message in Streamer Server Log:

#7(I)[2022-12-06 10:47:16,149]=18:47:16=00070{6216}&D671BB53& TCP RTSP Server: handling ANNOUNCE: streamid=control/13/0
#8(I)[2022-12-06 10:47:16,149]=18:47:16=00000{6216}&DC07F892& RtspHandshake::onRtspAnnounce : handling request: 7 289877937
…
#1(E)[2022-12-06 10:47:16,150]=18:47:16=00000{6216}&EE0C5B5F& Adapter 0 not found in the list of heads.
#2(E)[2022-12-06 10:47:16,150]=18:47:16=00000{6216}&3E35A65C& onConnectionRequest failed: Adapter not found.
#3(E)[2022-12-06 10:47:16,150]=18:47:16=00000{6216}&17C20B93& onConnectionRequest failed in Main server, resetting session state to initial

In my computer（normal），the message in Streamer Server Log

#3(I)[2022-12-01 08:04:51,570]=16:04:51=00064{19724}&D671BB53& TCP RTSP Server: handling ANNOUNCE: streamid=control/13/0
#4(I)[2022-12-01 08:04:51,570]=16:04:51=00000{19724}&DC07F892& RtspHandshake::onRtspAnnounce : handling request: 7 1859049732
#5(I)[2022-12-01 08:04:51,570]=16:04:51=00000{19724}&EBB67514& Is Valid SDP: 1
#6(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&19E2A18D& GetNvscConfigFromSdp
#7(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&483790F3& Successfully read 9 NvscClientConfig attributes from SDP.
#8(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&BC2AF0B4& Server is handling a request to establish a connection.
#9(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&4134E183& Head 0: DevName \.\DISPLAY1, DxAdapterOrdinal 0, DevString ‘Intel(R) Iris(R) Xe Graphics’. AttachedToDesktop MonitorConnected
#0(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&4134E183& Head 1: DevName \.\DISPLAY2, DxAdapterOrdinal 0, DevString ‘Intel(R) Iris(R) Xe Graphics’. MonitorConnected
#1(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&4134E183& Head 2: DevName \.\DISPLAY3, DxAdapterOrdinal 0, DevString ‘Intel(R) Iris(R) Xe Graphics’. MonitorConnected
#2(I)[2022-12-01 08:04:51,571]=16:04:51=00000{19724}&4134E183& Head 3: DevName \.\DISPLAY4, DxAdapterOrdinal 0, DevString ‘Intel(R) Iris(R) Xe Graphics’. MonitorConnected

It has a list of heads。

why the problem environment（GPU T4） has no heads?

CloudXR Version:3.2.2
GPU: T4, Pass-Through mode with VM.

CloudXR Server - SteamVR Log 2022-12-06 18.45.56.txt (922 Bytes)
CloudXR Server - SteamVR Log 2022-12-06 18.46.05.txt (1.6 KB)
Streamer Server Log 2022-12-06 18.46.05.txt (771.5 KB)

3

For the T4 system, are you attempting to steam from a virtualized instance (VMWare, AWS, etc)?

4

Thank you for reply，I have steam from KVM Instance. Now I have resolve it by updating the win10 to the latest version

5

Are you still running into this issue? If so, please provide as much detail as possible as to the server setup. Will is out this week, but can respond in the next week if you can give us more information on the physical setup.

I’ll note the two things that stand out in the server log, the first is just a bit odd, the second is the indicator of the underlying issue.

#1(M)[2022-12-06 10:46:09,895]=18:46:09=*00000*{6132}&F0E847F4&<AdapterContext> Initialized context for adapter 0: 3456 x 2160 @ 32.0 Hz

This seems strange to me. The remote head is set to not-quite-4k, at 32hz? I’ve not seen this before in logs, but Will is the expert for VM environment cases.

Then you will see:

#5(M)[2022-12-06 10:46:09,895]=18:46:09=*00000*{6132}&8BAD092A&<VideoEncoder> Initializing NvEnc10VideoEncoder. #6(M)[2022-12-06 10:46:09,895]=18:46:09=*00000*{6132}&2A6D6983&<NvEnc10VideoEnco> Using NvEnc header version 10.0. #7(E)[2022-12-06 10:46:09,995]=18:46:09=*00100*{6132}&19B2C542&<NvEnc10VideoEnco> Failed to open CUDA encoding session for checking perf support (NV_ENC_ERR_UNSUPPORTED_DEVICE). #8(W)[2022-12-06 10:46:09,995]=18:46:09=*00000*{6132}&DFA07824&<NvEnc10VideoEnco> Failed to get async encode support capability #9(W)[2022-12-06 10:46:09,997]=18:46:09=*00001*{6132}&5855CDE4&<Utils> Failed to get primary display ID, error code: -6

This is the real issue. So, again, Will is the expert here and I’ll wait for his knowledge, but there are two potential issues. One is a driver licensing issue, where I think certain server products need to be licensed for pass through usage.

But, if that’s not an issue, the next is there’s a known driver bug in identifying certain enterprise boards/chips, generally in VM environment. I’m not sure if there is a workaround in 3.2.2.

Again, I’m just stepping in as hadn’t seen a further response from you, could really use more details on your specific server setup to try and determine which issue it is, and whether we have something to try to get around it.

Thanks!

-d