Hi all, 

 

Up until about a week ago my UnRAID server worked exceptionally well, with very few crashes. Over the last week, unraid has been crashing multiple times a day. I did recently add a Google Coral USB device to my machine, and then 'beefed up' my home surveillance system (running Frigate as an NVR and using CompreFace and DoubleTake for motion detection/object+facial recognition. I'm hoping this is not the culprit... 

 

The other services I run that are "power hogs" are Plex and Tdarr. 

 

I did notice the other day in my syslog logs that I was getting some "out of memory" messages. I have 32 GB of RAM installed, but I do see on my dashboard that RAM at baseline is at about 30%, but under the "Docker" bar under "SYSTEM" in the dash, it says 63% (61.9 GiB). 

 

I've attached my diagnostics. I also have my syslogs downloaded and I'm happy to share them. The lines that caught my eye include: 

Multiple instances of "out_of_memory....":

```

Aug  3 10:20:25 10.0.1.36 kernel: Call Trace:
Aug  3 10:20:25 10.0.1.36 kernel: <TASK>
Aug  3 10:20:25 10.0.1.36 kernel: dump_stack_lvl+0x44/0x5c
Aug  3 10:20:25 10.0.1.36 kernel: dump_header+0x4a/0x211
Aug  3 10:20:25 10.0.1.36 kernel: oom_kill_process+0x80/0x111
Aug  3 10:20:25 10.0.1.36 kernel: out_of_memory+0x3b3/0x3e5

```

 

`Aug  5 06:08:52 10.0.1.36 root: Fix Common Problems: Error: Out Of Memory errors detected on your server`

 

 

cwunraid-diagnostics-20230806-0932.zip

Enable the syslog server and post that after a crash.

 

Im having the same issue I can ping the unraid server but can not SSH neither access the web ui 

this started after adding this config to Frigate 

 

Enable go2rtc: with 5 cameras 

this are my recording detectors and options 
 

###### Detectors ######
detectors:
  tensorrt:
    type: tensorrt
    device: 0 #This is the default, select the first GPU

model:
  path: /trt-models/yolov7-tiny-416.trt
  input_tensor: nchw
  input_pixel_format: rgb
  width: 416
  height: 416
  
ffmpeg:
  hwaccel_args: preset-nvidia-h265
  
objects:
  track:
    - person
    - bird
    - car
    - dog
    - motorcycle
    - bicycle
    - Package
    - Ups
    - Amazon
database:
  path: /config/frigate.db

###### Recording ######
snapshots:
  enabled: True
  timestamp: true
  bounding_box: true
  retain: 
    default: 30
record:
  enabled: True
  events:
    pre_capture: 5
    post_capture: 5
    retain:
      default: 30
      mode: active_objects
    objects:
      - person 
      - bird
      - car
      - dog
      - motorcycle
      - bicycle
      - Package
      - Ups
      - Amazon      

###### Detectors ######
detectors:
  tensorrt:
    type: tensorrt
    device: 0 #This is the default, select the first GPU

model:
  path: /trt-models/yolov7-tiny-416.trt
  input_tensor: nchw
  input_pixel_format: rgb
  width: 416
  height: 416
  
ffmpeg:
  hwaccel_args: preset-nvidia-h265
  
objects:
  track:
    - person
    - bird
    - car
    - dog
    - motorcycle
    - bicycle
    - Package
    - Ups
    - Amazon
database:
  path: /config/frigate.db

###### Birdseyeng ######
birdseye:
  # Optional: Enable birdseye view (default: shown below)
  enabled: True
  # Optional: Width of the output resolution (default: shown below)
  width: 1280
  # Optional: Height of the output resolution (default: shown below)
  height: 720
  # Optional: Encoding quality of the mpeg1 feed (default: shown below)
  # 1 is the highest quality, and 31 is the lowest. Lower quality feeds utilize less CPU resources.
  quality: 1
  # Optional: Mode of the view. Available options are: objects, motion, and continuous
  #   objects - cameras are included if they have had a tracked object within the last 30 seconds
  #   motion - cameras are included if motion was detected in the last 30 seconds
  #   continuous - all cameras are included always
  mode: objects

rtmp:
  enabled: false

live:
  height: 640
  quality: 1
###### MQTT and Connectors ######
mqtt:
  enabled: False

 

unraidserver-diagnostics-20230901-0759.zip unraidserver-syslog-20230901-1200.zip

Anyone of y'all found this issue? I'm starting to think that this issue is also happening on my main unraid running frigate.