Nest hickup every * minutes

[PrutsMeneer]

Hi! I’m running into a very consistent issue that’s been driving me crazy for the past two weeks: parts of my Nest recordings are missing because the video feed briefly freezes every few minutes.

What happens

Using Go2RTC with Nest, the Nest video feed freezes for a couple of seconds and then recovers. This repeats very consistently, for example, exactly every 5 minutes and 24 seconds.

Each time it happens, I see the following log line:
debug | [streams] retry=0 to url=nest:?client_id=**********&client_secret=****&device_id=*****&project_id=****&protocols=WEB_RTC&refresh_token=***

Environment
• go2rtc version=1.9.14 platform=linux/amd64 revision=b5948cf

api:
  listen: :1984

streams:
  voor_tuin_nest:
    - nest:?client_id=****&client_secret=****&device_id=***&protocols=WEB_RTC&refresh_token=***

  achter_tuin_nest:
    - nest:?client_id=****&client_secret=****&device_id=***&protocols=WEB_RTC&refresh_token=***

rtsp:
  listen: :8554

log:
  level: info  # default level
  api: debug
  exec: debug
  rtsp: debug
  streams: trace
  webrtc: debug

[PrutsMeneer]

Hi @AlexxIT,

I think i solved the problem. I need to admit: with some help of ChatGPT (Codex),
But I've been running this for about a week now and the problem seems to be solved; no longer i did have any reconnect!

What has been done:

• make Nest token refresh deterministic by storing credentials in API and calling OAuth refresh directly
• make ExtendStream retry behavior more robust for transient 401/409/429 responses
• keep extend timer loop running after transient extend errors to avoid unnecessary stream reconnects

Why
When stream extension fails once, the stream can expire and streams reconnect starts ([streams] retry=...). These changes reduce avoidable drops by improving retry and refresh behavior.

I created a PR
PR