I have created an EC2 using the 3.0 ami, followed the instructions, installing Steam, then SteamVR, then the SDK and running the firewall .bat, then installed the Client on my Oculus Quest 2 using adb per documentation and pushed an -s [ip address] to /sdcard/CloudXRLaunchOptions.txt (also using adb).
When I launch SteamVR and try to connect using the client app, the HMD cannot get past the rotating loading icon and the CXR icon does not appear in the SteamVR server app. CloudXRRemoteHMD is listed as a SteamVR addon and turned on.
Port configs are per default settings, running via a static elastic ip.
No logs can be found on the OQ2 as the /sdcard/CloudXR folder does not exist after running the client. Developer mode is enabled on the headset.
Below is the log from the server:
NVIDIA CloudXR v3.0 (CL# 30264573), built on Aug 4 2021 20:46:27.
[##] Cleaning up older files.
[##] VisionServerDriver::Init()
[##] CloudXR single instance mutex created, first one.
[##] VisionServerDriver::RunFrame() – creating VisionHmdDriver
[##] VisionHmdDriver()
[##] VSync to photons: 0.0200.
[##] Adapter 0: Description=‘NVIDIA Tesla T4’
[##] Direct3D setup done.
[##] Send audio: Enabled, Receive audio: Disabled, Foveation: Two pass
[##] Starting FrameServerDX11.
[##] Retry loading stream server dll from cxr directory
[##] Stream Server: Net Stream Server Instance Created
[##] Using an app provided render device for video processing.
[##] Stream Server: CloudXR video streams created
[##] Stream Server: Input stream created
[##] Stream Server: Audio send stream created
[##] Stream Server: streaming instance started - waiting for a client…
[##] VisionHmdDriver::VisionHmdDriver succeeded.
Whenever a CXR Client connects to the server, there will be several logs of connection attempts for each type of service (video/audio/etc). The line item: [##] Stream Server: streaming instance started - waiting for a client suggest that the CXR client isn’t even making it to the server. There’s likely something upstream that is blocked on the network that is preventing CXR connections to begin with. Have you double-checked your AWS Security Group settings within the VPC the EC2 instance is deployed in? Also, have you verified your outbound network supports ephemeral ports if outbound firewalls on your network are enabled?
UPDATE1: I ran an AWS reachability analysis via port 48010 to the EC2 from my aws account internet gateway and it was successful both ways.
UPDATE2: I’m also able to ping my home wifi router IPv4 from the AWS VM via cmd/ping.
UPDATE3: Turning off AWS VM Windows firewall AND home router firewall has not solved the issue.
Not sure how to check the ephemeral ports, could you provide a walkthrough pls?
Also, is there a reason I cannot find sdcard/CloudXR/ on my HMD? (i.e. looking for the client logs) I’ve added -v for verbose logging to CloudXRLaunchOptions.txt, but still no logs folder or files on the OQ2…
UPDATE4: tried the following with SteamVR running on the EC2 instance (x.x.x.x being the public vm ip): nmap -p 47998-48010 x.x.x.x result:
Starting Nmap 7.92 ( https://nmap.org ) at yyyy-mm-dd hh:mm xxx Daylight Time
pcap_activate(\Device\NPF_xxx) WARNING: Unknown NdisMedium value 19, defaulting to DLT_EN10MB.
refresh_hostbatch: Failed to determine dst MAC address for target x.x.x.x
Note: Host seems down. If it is really up, but blocking our ping probes, try -Pn
Nmap done: 1 IP address (0 hosts up) scanned in 1.19 seconds
result using -Pn:
Starting Nmap 7.92 ( https://nmap.org ) at yyyy-mm-dd hh:mm xxx Daylight Time
pcap_activate(\Device\NPF_xxx) WARNING: Unknown NdisMedium value 19, defaulting to DLT_EN10MB.
refresh_hostbatch: Failed to determine dst MAC address for target x.x.x.x
Nmap done: 1 IP address (0 hosts up) scanned in 1.22 seconds
