[Support]: Frigate doesn't start on TrueNAS after update

[acelot]

Describe the problem you are having

Hello everyone!

I've been using Frigate for over a year now. The service was configured and running successfully until today, when I decided to update the version from 0.12.1 to 0.13.2. The service just hangs after [INFO] Starting go2rtc healthcheck service... (see logs).

I'm using:

• TrueNAS: 23.10.1.1
• Frigate Helm Chart: 1.2.6
• Frigate Docker Image: ghcr.io/blakeblackshear/frigate:0.13.2-tensorrt

Version

0.13.2

Frigate config file

mqtt:
  enabled: true
  host: 172.16.0.17
  port: 1883

cameras:
  dacha:
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=0
          roles:
            - detect
            - record
    detect:
      enabled: true
      fps: 5
      width: 1280
      height: 720
    zones:
      grill:
        coordinates: 139,720,45,720,0,623,0,271,263,221,810,408
      porch:
        coordinates: 1166,535,1237,306,865,214,836,412
      terrace:
        coordinates: 1095,720,1166,528,822,410,146,720
      yard:
        coordinates: 265,229,0,281,0,173,497,121,848,197,818,410
      sauna:
        coordinates: 1175,383,1223,221,1162,179,1074,173,1046,342

detect:
  stationary:
    interval: 0
    threshold: 50

objects:
  track:
    - person
    - cat
    - dog
  filters:
    person:
      threshold: 0.8
    cat:
      threshold: 0.8
    dog:
      threshold: 0.8
  mask: 545,131,1237,300,1280,145,875,64,756,0,0,0,0,180

record:
  enabled: true
  retain:
    days: 3
    mode: motion
  events:
    retain:
      default: 14
      mode: active_objects

ffmpeg:
  hwaccel_args: preset-nvidia-h264

detectors:
 tensorrt:
   type: tensorrt

model:
 path: /trt-models/yolov7-tiny-416.trt
 input_tensor: nchw
 input_pixel_format: rgb
 width: 416
 height: 416

ui:
  live_mode: mse
  timezone: Asia/Novosibirsk

Relevant log output

2024-02-21 18:51:15.179221+07:00s6-rc: info: service s6rc-fdholder: starting
2024-02-21 18:51:15.179320+07:00s6-rc: info: service s6rc-oneshot-runner: starting
2024-02-21 18:51:15.182428+07:00s6-rc: info: service s6rc-oneshot-runner successfully started
2024-02-21 18:51:15.182737+07:00s6-rc: info: service fix-attrs: starting
2024-02-21 18:51:15.184075+07:00s6-rc: info: service s6rc-fdholder successfully started
2024-02-21 18:51:15.187781+07:00s6-rc: info: service fix-attrs successfully started
2024-02-21 18:51:15.187929+07:00s6-rc: info: service legacy-cont-init: starting
2024-02-21 18:51:15.191497+07:00s6-rc: info: service legacy-cont-init successfully started
2024-02-21 18:51:15.191556+07:00s6-rc: info: service trt-model-prepare: starting
2024-02-21 18:51:15.191642+07:00s6-rc: info: service log-prepare: starting
2024-02-21 18:51:15.204515+07:00/etc/s6-overlay/s6-rc.d/trt-model-prepare/run: line 22: YOLO_MODELS: unbound variable
2024-02-21 18:51:15.204690+07:00s6-rc: warning: unable to start service trt-model-prepare: command exited 1
2024-02-21 18:51:15.207001+07:00s6-rc: info: service log-prepare successfully started
2024-02-21 18:51:15.207127+07:00s6-rc: info: service nginx-log: starting
2024-02-21 18:51:15.207277+07:00s6-rc: info: service go2rtc-log: starting
2024-02-21 18:51:15.207394+07:00s6-rc: info: service frigate-log: starting
2024-02-21 18:51:15.208585+07:00s6-rc: info: service nginx-log successfully started
2024-02-21 18:51:15.208748+07:00s6-rc: info: service go2rtc-log successfully started
2024-02-21 18:51:15.208836+07:00s6-rc: info: service go2rtc: starting
2024-02-21 18:51:15.208997+07:00s6-rc: info: service frigate-log successfully started
2024-02-21 18:51:15.210325+07:00s6-rc: info: service go2rtc successfully started
2024-02-21 18:51:15.210435+07:00s6-rc: info: service go2rtc-healthcheck: starting
2024-02-21 18:51:15.212061+07:00s6-rc: info: service go2rtc-healthcheck successfully started
2024-02-21 18:51:15.265353+07:002024-02-21 11:51:15.263317779  [INFO] Preparing new go2rtc config...
2024-02-21 18:51:15.701500+07:002024-02-21 11:51:15.701413100  [INFO] Starting go2rtc...
2024-02-21 18:51:15.767242+07:002024-02-21 11:51:15.767130232  11:51:15.767 INF go2rtc version 1.8.4 linux/amd64
2024-02-21 18:51:15.767571+07:002024-02-21 11:51:15.767514350  11:51:15.767 INF [api] listen addr=:1984
2024-02-21 18:51:15.767596+07:002024-02-21 11:51:15.767581646  11:51:15.767 INF [rtsp] listen addr=:8554
2024-02-21 18:51:15.767708+07:002024-02-21 11:51:15.767678766  11:51:15.767 INF [webrtc] listen addr=:8555
2024-02-21 18:51:25.215749+07:002024-02-21 11:51:25.215684068  [INFO] Starting go2rtc healthcheck service...
2024-02-21 18:56:20.687821+07:00s6-rc: info: service nginx-log: stopping
2024-02-21 18:56:20.687929+07:00s6-rc: info: service go2rtc-healthcheck: stopping
2024-02-21 18:56:20.688061+07:00s6-rc: info: service frigate-log: stopping
2024-02-21 18:56:20.692317+07:002024-02-21 11:56:20.692252227  [INFO] The go2rtc-healthcheck service exited with code 256 (by signal 15)
2024-02-21 18:56:20.692865+07:00s6-rc: info: service go2rtc-healthcheck successfully stopped
2024-02-21 18:56:20.692986+07:00s6-rc: info: service go2rtc: stopping
2024-02-21 18:56:20.693753+07:002024-02-21 11:56:20.693708692  exit with signal: terminated
2024-02-21 18:56:20.697274+07:002024-02-21 11:56:20.697218057  [INFO] The go2rtc service exited with code 0 (by signal 0)
2024-02-21 18:56:20.697835+07:00s6-rc: info: service go2rtc successfully stopped
2024-02-21 18:56:20.697934+07:00s6-rc: info: service go2rtc-log: stopping

FFprobe output from your camera

N/A

Frigate stats

No response

Operating system

Other Linux

Install method

Docker CLI

Coral version

CPU (no coral)

Network connection

Wired

Camera make and model

N/A

Any other information that may be helpful

No response

[acelot]

My TensorRT models are already baked, so I disabled models preparing by passing YOLO_MODELS="" to env variables

[NickM-27]

You can not use your existing trt models, this is covered in the release notes, they must be regenerated and your must update your config according to the docs

[acelot]

OK, I'll try to regenerate, it's a very long process. However there is a problem with K8S Backoff timeouts, need to fix helm charts.

[acelot]

Nevermind, I've set YOLO_MODELS=yolov7-tiny-416 env variable and model was created succesfully!