Skip to main content
New Participant
May 14, 2026
Question

Link connection get freezed

  • May 14, 2026
  • 23 replies
  • 1994 views

ENG: 

Since the major OS update, the Link cable connection has been failing and randomly disconnecting for a moment, freezing the VR image and making it impossible to reconnect without fully restarting the PC.

I mainly use the headset for sim racing (iRacing, LMU), and this happens quite frequently.

My headset settings are not aggressive. After several attempts and configuration changes, it can sometimes last a few hours without issues, but then it suddenly freezes again. The game and the headset itself continue working — I can still see on the monitor that the headset movements are being tracked in-game — but I cannot see anything through the headset anymore.

I do not currently use an official Link cable (although I have ordered the official €80 cable to test whether it solves the issue), but I never experienced problems like this before. Everything started after the major Meta OS update.

Since then, I have also experienced zombie processes that would not close properly, causing the image to run at 2 FPS and leading to crashes. Now this. 

SPA:

Desde la gran actualización donde se cambia el SO, el cable link falla y se desconecta por un momento, congelando la imagen de VR y haciendo imposible volver a conectar sin reiniciar el PC entero.

Suelo jugar a Simracing con ellos (iRacing, LMU) y me pasa con bastante asiduidad. 

La configuración de las gafas no es agresiva. Después de varios intentos y cambios ha aguantado unas horas sin quejarse hasta que de repente se congela. El juego y las gafas siguen funcionando, incluso en la pantalla sigo viendo como las gafas están aun provocando movimientos en el juego, pero no puedo ver nada desde ellas.

No tengo un cable Link oficial (he pedido el cable de 80 euros para ver si con esto se arregla) pero nunca había tenido problemas así, solo desde la gran actualización del SO de Meta. Desde entonces, procesos zombieficados que no se apagaban, provocando que la imagen fuera a 2 Fps y crasheos. Y ahora esto...

PD: I tried a clean install of Nvidia Drivers + Turn off Meta Notifications. After some days testing, Meta Link crashed again and freezes. Doesnt work

23 replies

New Participant
August 11, 2026

Adding detailed diagnostics, because our rig hits exactly the same issue (iRacing, Quest 3 via USB-C cable Link, RTX 5070 Ti, Windows 11) and after pulling the runtime logs apart I can narrow down where it dies.

 

Short version: in our case the trigger is the Meta PC software v206. Rolling the PC app back to v205 stopped the freezes immediately. GPU driver, bitrate, codec and USB were ruled out with data.

 

Timeline (same NVIDIA driver throughout, H.264/AVC, fixed bitrate, dynamic bitrate disabled):

- PC software v205 (205.0.0.167.543): multi-hour sessions, zero freezes (longest 8.4 h).

- v206 (206.0.0.103.432) auto-installed itself. First freeze the NEXT day. At least 7 hard freezes over the following 6 days, stream survival shrinking from hours to 10-15 minutes.

- Rolled back to v205: clean sessions again.

- v206 force-updated itself again: froze after 15 minutes in the very next session.

- Rolled back to v205 once more.

 

What the logs show at every freeze:

1. The headset-side link client keeps logging decoded frames at a steady 90 fps for at least a minute AFTER the image freezes. So encoder, USB transport and headset decoder are all still alive - the server is just delivering the same frame forever. That is why the frozen image "sticks" to your head: ASW keeps reprojecting the last frame it got.

2. The PC service's once-per-minute telemetry heartbeat stops abruptly, with not a single error line in the service log.

3. 15-40 seconds later, every single time: {WARNING} [LinkClient] Watchdog triggered: Dash home-button forwarding temporarily disabled - the server's IPC path stops responding.

4. Once, at shutdown, the service logged "AppFocusObserver thread took longer than 50s: 7175379ms" - a service thread had been hung for about two hours.

5. Windows event logs are clean at every freeze timestamp: no GPU driver resets, no USB errors. The USB keepalive ping/pong continues for minutes after the freeze.

 

Ruled out with data on our rig: NVIDIA driver version (unchanged across healthy and broken periods - and note the AMD report above, which also points away from NVIDIA), encoder backup (zero dropped-frame events at freeze time), bitrate (froze at reduced bitrate during the statistically cleanest session), H.265 (we were on H.264 the whole time, so the codec workaround above cannot be the whole story for this variant), USB port (chipset and CPU-direct both tried), USB power management, GPU overclock (froze at stock clocks).

 

One more v206-specific observation: it spams "Server (highwind_dap_server) not found" plus "InitClient() failed" retries every 10 seconds (more than 1300 per session). That code path does not exist in v205 (zero occurrences). It looks like a half-finished transport rework, and the freeze behaves like a deadlock in the new frame-routing/compositor path.

 

Conclusion: for the cable-Link variant described here, this is a v206 PC software bug, not a GPU driver or cable issue. Since the LIVE channel force-updates the PC app, the only reliable workaround for us is staying on v205 and blocking the updater - which is obviously not sustainable.

 

Meta: full logs available on request. Please escalate this to the PC software team.

New Participant
September 3, 2026

Hi nexudx,

you mentioned rolling Meta Horizon Link back from v206 to 205.0.0.167.543 twice.

I'm currently dealing with the same OpenXR regression on v207.

Could you share where you got the v205 files, or upload your v205 Support folder/archive? Even just the CONFIG : {...} line from the OculusSetup.log from that v205 installation would be extremely useful.

 

Thanks!