April 21, 20251 yr 2 hours ago, yayitazale said: Can you post the error? Sure thing. For fun I put the shim to 16GB (--shm-size=16gb); same error. Error: failed to register layer: write /usr/local/cuda/targets/x86_64-linux/lib/libcublasLt.so.12.1.0.26: no space left on device CUDA would point to the graphics card. My config currently has all the related nvidia gpu related settings defined in the container. Edited April 21, 20251 yr by repomanz
April 22, 20251 yr Author 9 hours ago, repomanz said: Sure thing. For fun I put the shim to 16GB (--shm-size=16gb); same error. Error: failed to register layer: write /usr/local/cuda/targets/x86_64-linux/lib/libcublasLt.so.12.1.0.26: no space left on device CUDA would point to the graphics card. My config currently has all the related nvidia gpu related settings defined in the container. I suggest you to open a issue https://github.com/blakeblackshear/frigate/issues I cannot help you with this error.
April 23, 20251 yr GPU Goes to 100%... Not sure where to start with this as the system runs 'normally' for several days or weeks then the GPU goes to 100% load... Using the GPU Statistics plugin, the GPU is shown as Vendor: GPU, Driver: AMDGPU, Device: Radeon Vega Series... Once the GPU goes to 100%, I stop all docker and VM instances but that does not help. The only thing I have configured to use the GPU is Frigate (a Docker application). Rebooting Unraid brings the GPU usage back to the usual 3-5% Has anyone seen this behaviour and/or know of any utility to help identify the process consuming the GPU? Thanks...
April 23, 20251 yr On 4/22/2025 at 12:47 AM, yayitazale said: I suggest you to open a issue https://github.com/blakeblackshear/frigate/issues I cannot help you with this error. To close the loop here. I ended up opening a ticket with the frigate devs. The issue was my total docker size was the problem. Increased it by 10gb; issue resolved.
May 30, 20251 yr Just FYI, I started having loads of issues since 0.16 betas - Turns out I also made changes to my extra parameters, and I increased my shm sizing but that same time I somehow reduced my tmpfs-size=100000000 which equals to only 100MB - Maybe it was always set to that from a really old Template?But often I would just not get recordings anymore, I now gave it 25GB as I have plenty of RAM to spare and all my issues have disappeared.So just double check your Extra Params
June 2, 20251 yr Author On 5/30/2025 at 11:00 PM, valiente said:Just FYI, I started having loads of issues since 0.16 betas - Turns out I also made changes to my extra parameters, and I increased my shm sizing but that same time I somehow reduced my tmpfs-size=100000000 which equals to only 100MB - Maybe it was always set to that from a really old Template?But often I would just not get recordings anymore, I now gave it 25GB as I have plenty of RAM to spare and all my issues have disappeared.So just double check your Extra ParamsDefault in the template:--shm-size=256m --mount type=tmpfs,target=/tmp/cache,tmpfs-size=1000000000 --restart unless-stopped
June 11, 20251 yr So I was previously running a frigate instance with a quadro p1000 and a reolink doorbell with no problems but lost the config when I changed to an rtx a4000. I tried to generate a new config and it seems to be in a restart loop. anyone can help?config:mqtt: host: 192.168.50.5 user: frigatemqtt password: xxxxxxxxxxxdetectors: tensorrt: type: tensorrt device: 0 model: path: /config/model_cache/tensorrt/8.5.3/yolov7-tiny-416.trt input_tensor: nchw input_shape: [1, 3, 416, 416]go2rtc: streams: doorbell: - "ffmpeg:http://192.168.50.19/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=xxxxxxxxxxxxxx&password=xxxxxxxxxxx#video=copy#audio=copy#audio=opus" - rtsp://192.168.50.19/Preview_01_sub doorbell_sub: - "ffmpeg:http://192.168.50.19/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=xxxxxxxxxxxxx&password=xxxxxxxxxxxxx"cameras: doorbell: ffmpeg: inputs: - path: rtsp://127.0.0.1:8554/doorbell input_args: preset-rtsp-restream roles: - record - path: rtsp://127.0.0.1:8554/doorbell_sub input_args: preset-rtsp-restream roles: - detect detect: width: 416 height: 416 fps: 5 snapshots: enabled: true timestamp: true bounding_box: true retain: default: 2 record: enabled: true retain: days: 7 mode: motionlogss6-rc: info: service s6rc-fdholder: startings6-rc: info: service s6rc-oneshot-runner: startings6-rc: info: service s6rc-oneshot-runner successfully starteds6-rc: info: service fix-attrs: startings6-rc: info: service s6rc-fdholder successfully starteds6-rc: info: service fix-attrs successfully starteds6-rc: info: service legacy-cont-init: startings6-rc: info: service legacy-cont-init successfully starteds6-rc: info: service trt-model-prepare: startings6-rc: info: service log-prepare: startings6-rc: info: service log-prepare successfully starteds6-rc: info: service nginx-log: startings6-rc: info: service go2rtc-log: startings6-rc: info: service frigate-log: startings6-rc: info: service certsync-log: startings6-rc: info: service trt-model-prepare successfully starteds6-rc: info: service nginx-log successfully starteds6-rc: info: service frigate-log successfully starteds6-rc: info: service go2rtc-log successfully starteds6-rc: info: service go2rtc: startings6-rc: info: service certsync-log successfully starteds6-rc: info: service go2rtc successfully starteds6-rc: info: service go2rtc-healthcheck: startings6-rc: info: service frigate: startings6-rc: info: service go2rtc-healthcheck successfully starteds6-rc: info: service frigate successfully starteds6-rc: info: service nginx: startings6-rc: info: service nginx successfully starteds6-rc: info: service certsync: startings6-rc: info: service certsync successfully starteds6-rc: info: service legacy-services: startings6-rc: info: service legacy-services successfully started2025-06-11 00:56:16.297994515 [2025-06-11 00:56:16] peewee.sqliteq INFO : writer received shutdown request, exiting.2025-06-11 00:56:16.386250666 [INFO] Service Frigate exited with code 0 (by signal 0)2025-06-11 00:56:16.388627480 exit with signal: terminated2025-06-11 00:56:16.394064340 [INFO] The go2rtc service exited with code 0 (by signal 0)No models to convert.2025-06-11 00:56:21.020697518 [INFO] Preparing new go2rtc config...2025-06-11 00:56:21.021912341 [INFO] Preparing Frigate...2025-06-11 00:56:21.023507578 [INFO] Starting NGINX...2025-06-11 00:56:21.097524499 [INFO] Starting Frigate...2025-06-11 00:56:21.099819157 [INFO] Starting certsync...2025-06-11 00:56:21.174543616 127.0.0.1 - - [11/Jun/2025:00:56:21 +0100] "" 400 0 "-" "-" "-"2025-06-11 00:56:21.339580567 [INFO] Starting go2rtc...2025-06-11 00:56:21.416299222 00:56:21.416 INF go2rtc platform=linux/amd64 revision=b2399f3 version=1.9.22025-06-11 00:56:21.416302448 00:56:21.416 INF config path=/dev/shm/go2rtc.yaml2025-06-11 00:56:21.416662613 00:56:21.416 INF [api] listen addr=:19842025-06-11 00:56:21.416867524 00:56:21.416 INF [rtsp] listen addr=:85542025-06-11 00:56:21.416989075 00:56:21.416 INF [webrtc] listen addr=:8555/tcp2025-06-11 00:56:22.834686221 [2025-06-11 00:56:22] frigate.util.config INFO : Checking if frigate config needs migration...2025-06-11 00:56:22.843014055 [2025-06-11 00:56:22] frigate.util.config INFO : frigate config does not need migration...2025-06-11 00:56:23.878169248 [2025-06-11 00:56:23] frigate.util.services INFO : Automatically detected nvidia hwaccel for video decoding2025-06-11 00:56:23.926827426 [2025-06-11 00:56:23] frigate.app INFO : Starting Frigate (0.15.0-6cb5cfb)2025-06-11 00:56:23.932025140 [2025-06-11 00:56:23] peewee_migrate.logs INFO : Starting migrations2025-06-11 00:56:23.932179464 [2025-06-11 00:56:23] peewee_migrate.logs INFO : There is nothing to migrate2025-06-11 00:56:23.954137113 [2025-06-11 00:56:23] frigate.app INFO : Recording process started: 4732025-06-11 00:56:23.955190679 [2025-06-11 00:56:23] frigate.app INFO : Review process started: 4822025-06-11 00:56:23.957056882 [2025-06-11 00:56:23] frigate.app INFO : go2rtc process pid: 1052025-06-11 00:56:23.972708176 [2025-06-11 00:56:23] detector.tensorrt INFO : Starting detection process: 4892025-06-11 00:56:23.988742370 [2025-06-11 00:56:23] frigate.app INFO : Output process started: 5062025-06-11 00:56:24.007124332 [2025-06-11 00:56:24] frigate.app INFO : Camera processor started for doorbell: 5232025-06-11 00:56:24.026536536 [2025-06-11 00:56:24] frigate.app INFO : Capture process started for doorbell: 5372025-06-11 00:56:24.122939122 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : Loaded engine size: 29 MiB2025-06-11 00:56:24.214014396 [2025-06-11 00:56:24] frigate.api.fastapi_app INFO : Starting FastAPI app2025-06-11 00:56:24.276486110 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +6, GPU +8, now: CPU 187, GPU 208 (MiB)2025-06-11 00:56:24.288052217 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuDNN: CPU +2, GPU +10, now: CPU 189, GPU 218 (MiB)2025-06-11 00:56:24.293538561 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +28, now: CPU 0, GPU 28 (MiB)2025-06-11 00:56:24.294846411 [2025-06-11 00:56:24] frigate.api.fastapi_app INFO : FastAPI started2025-06-11 00:56:24.306690979 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +1, GPU +8, now: CPU 160, GPU 210 (MiB)2025-06-11 00:56:24.306717700 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuDNN: CPU +0, GPU +8, now: CPU 160, GPU 218 (MiB)2025-06-11 00:56:24.309930656 [2025-06-11 00:56:24] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +136, now: CPU 0, GPU 164 (MiB)2025-06-11 00:56:31.022570692 [INFO] Starting go2rtc healthcheck service...2025-06-11 00:56:33.941571117 Fatal Python error: Segmentation fault2025-06-11 00:56:33.941574202 2025-06-11 00:56:33.941575204 Thread 0x00001500d26ad6c0 (most recent call first):2025-06-11 00:56:33.941604240 File "/usr/lib/python3.9/threading.py", line 312 in wait2025-06-11 00:56:33.941643334 File "/usr/lib/python3.9/multiprocessing/queues.py", line 233 in _feed2025-06-11 00:56:33.941677299 File "/usr/lib/python3.9/threading.py", line 892 in run2025-06-11 00:56:33.941718357 File "/usr/lib/python3.9/threading.py", line 954 in bootstrapinner2025-06-11 00:56:33.941755127 File "/usr/lib/python3.9/threading.py", line 912 in _bootstrap2025-06-11 00:56:33.941755668 2025-06-11 00:56:33.941762150 Current thread 0x00001501a00eb740 (most recent call first):2025-06-11 00:56:33.941806755 File "/opt/frigate/frigate/detectors/plugins/tensorrt.py", line 193 in <listcomp>2025-06-11 00:56:33.941854797 File "/opt/frigate/frigate/detectors/plugins/tensorrt.py", line 192 in doinference2025-06-11 00:56:33.941903900 File "/opt/frigate/frigate/detectors/plugins/tensorrt.py", line 309 in detect_raw2025-06-11 00:56:33.941945529 File "/opt/frigate/frigate/object_detection.py", line 95 in detect_raw2025-06-11 00:56:33.941986537 File "/opt/frigate/frigate/object_detection.py", line 145 in run_detector2025-06-11 00:56:33.942027215 File "/usr/lib/python3.9/multiprocessing/process.py", line 108 in run2025-06-11 00:56:33.942068453 File "/opt/frigate/frigate/util/process.py", line 41 in run_wrapper2025-06-11 00:56:33.942113900 File "/usr/lib/python3.9/multiprocessing/process.py", line 315 in _bootstrap2025-06-11 00:56:33.942161470 File "/usr/lib/python3.9/multiprocessing/popen_fork.py", line 71 in _launch2025-06-11 00:56:33.942357403 File "/usr/lib/python3.9/multiprocessing/popen_fork.py", line 19 in init2025-06-11 00:56:33.942404282 File "/usr/lib/python3.9/multiprocessing/context.py", line 277 in _Popen2025-06-11 00:56:33.942449879 File "/usr/lib/python3.9/multiprocessing/context.py", line 224 in _Popen2025-06-11 00:56:33.942494564 File "/usr/lib/python3.9/multiprocessing/process.py", line 121 in start2025-06-11 00:56:33.942532366 File "/opt/frigate/frigate/util/process.py", line 30 in start2025-06-11 00:56:33.942582121 File "/opt/frigate/frigate/object_detection.py", line 204 in start_or_restart2025-06-11 00:56:33.942666281 File "/opt/frigate/frigate/object_detection.py", line 172 in init2025-06-11 00:56:33.942727197 File "/opt/frigate/frigate/app.py", line 350 in start_detectors2025-06-11 00:56:33.942750802 File "/opt/frigate/frigate/app.py", line 598 in start2025-06-11 00:56:33.942805877 File "/opt/frigate/frigate/__main__.py", line 60 in main2025-06-11 00:56:33.942843749 File "/opt/frigate/frigate/__main__.py", line 64 in <module>2025-06-11 00:56:33.942883044 File "/usr/lib/python3.9/runpy.py", line 87 in runcode2025-06-11 00:56:33.942926957 File "/usr/lib/python3.9/runpy.py", line 197 in runmodule_as_main2025-06-11 00:56:44.213061641 [2025-06-11 00:56:44] frigate.watchdog INFO : Detection appears to be stuck. Restarting detection process...2025-06-11 00:56:44.229847465 [2025-06-11 00:56:44] detector.tensorrt INFO : Starting detection process: 6682025-06-11 00:56:44.390762970 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : Loaded engine size: 29 MiB2025-06-11 00:56:44.572463613 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +6, GPU +8, now: CPU 191, GPU 413 (MiB)2025-06-11 00:56:44.583668953 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuDNN: CPU +2, GPU +10, now: CPU 193, GPU 423 (MiB)2025-06-11 00:56:44.595719303 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +28, now: CPU 0, GPU 28 (MiB)2025-06-11 00:56:44.601362596 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +0, GPU +8, now: CPU 164, GPU 415 (MiB)2025-06-11 00:56:44.601382824 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] Init cuDNN: CPU +0, GPU +8, now: CPU 164, GPU 423 (MiB)2025-06-11 00:56:44.614089143 [2025-06-11 00:56:44] frigate.detectors.plugins.tensorrt INFO : [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +136, now: CPU 0, GPU 164 (MiB)2025-06-11 00:56:44.616548524 Fatal Python error: Segmentation fault2025-06-11 00:56:44.616549937 2025-06-11 00:56:44.616550769 Thread 0x0000150088dbf6c0 (most recent call first):2025-06-11 00:56:44.616588741 File "/usr/lib/python3.9/threading.py", line 312 in wait2025-06-11 00:56:44.616651049 File "/usr/lib/python3.9/multiprocessing/queues.py", line 233 in _feed2025-06-11 00:56:44.616677951 File "/usr/lib/python3.9/threading.py", line 892 in run2025-06-11 00:56:44.616720512 File "/usr/lib/python3.9/threading.py", line 954 in bootstrapinner2025-06-11 00:56:44.616783201 File "/usr/lib/python3.9/threading.py", line 912 in _bootstrap2025-06-11 00:56:44.616786978 2025-06-11 00:56:44.616788140 Current thread 0x000015008a3f16c0 (most recent call first):2025-06-11 00:56:44.616801426 File "/opt/frigate/frigate/detectors/plugins/tensorrt.py", line 193 in <listcomp>2025-06-11 00:56:44.616867912 File "/opt/frigate/frigate/detectors/plugins/tensorrt.py", line 192 in doinference2025-06-11 00:56:44.616897959 File "/opt/frigate/frigate/detectors/plugins/tensorrt.py", line 309 in detect_raw2025-06-11 00:56:44.616934249 File "/opt/frigate/frigate/object_detection.py", line 95 in detect_raw2025-06-11 00:56:44.616971830 File "/opt/frigate/frigate/object_detection.py", line 145 in run_detector2025-06-11 00:56:44.617010714 File "/usr/lib/python3.9/multiprocessing/process.py", line 108 in run2025-06-11 00:56:44.617050931 File "/opt/frigate/frigate/util/process.py", line 41 in run_wrapper2025-06-11 00:56:44.617100164 File "/usr/lib/python3.9/multiprocessing/process.py", line 315 in _bootstrap2025-06-11 00:56:44.617155340 File "/usr/lib/python3.9/multiprocessing/popen_fork.py", line 71 in _launch2025-06-11 00:56:44.617209432 File "/usr/lib/python3.9/multiprocessing/popen_fork.py", line 19 in init2025-06-11 00:56:44.617253697 File "/usr/lib/python3.9/multiprocessing/context.py", line 277 in _Popen2025-06-11 00:56:44.617298442 File "/usr/lib/python3.9/multiprocessing/context.py", line 224 in _Popen2025-06-11 00:56:44.617341614 File "/usr/lib/python3.9/multiprocessing/process.py", line 121 in start2025-06-11 00:56:44.617384786 File "/opt/frigate/frigate/util/process.py", line 30 in start2025-06-11 00:56:44.617439430 File "/opt/frigate/frigate/object_detection.py", line 204 in start_or_restart2025-06-11 00:56:44.617487191 File "/opt/frigate/frigate/watchdog.py", line 33 in run2025-06-11 00:56:44.617537748 File "/usr/lib/python3.9/threading.py", line 954 in bootstrapinner2025-06-11 00:56:44.617578325 File "/usr/lib/python3.9/threading.py", line 912 in _bootstrap2025-06-11 00:56:54.226051226 [2025-06-11 00:56:54] frigate.watchdog INFO : Detection appears to have stopped. Exiting Frigate...s6-rc: info: service legacy-services: stoppings6-rc: info: service legacy-services successfully stoppeds6-rc: info: service go2rtc-healthcheck: stoppings6-rc: info: service certsync: stopping2025-06-11 00:56:54.236261944 [INFO] The go2rtc-healthcheck service exited with code 256 (by signal 15)s6-rc: info: service go2rtc-healthcheck successfully stopped2025-06-11 00:56:54.237915592 [INFO] Service CERTSYNC exited with code 256 (by signal 15)s6-rc: info: service certsync successfully stoppeds6-rc: info: service nginx: stoppings6-rc: info: service certsync-log: stoppings6-rc: info: service certsync-log successfully stopped2025-06-11 00:56:54.293940894 [INFO] Service NGINX exited with code 0 (by signal 0)s6-rc: info: service nginx successfully stoppeds6-rc: info: service nginx-log: stoppings6-rc: info: service frigate: stoppings6-rc: info: service nginx-log successfully stopped2025-06-11 00:56:54.458183161 [2025-06-11 00:56:54] frigate.app INFO : Stopping...2025-06-11 00:56:54.459200988 [2025-06-11 00:56:54] frigate.stats.emitter INFO : Exiting stats emitter...2025-06-11 00:56:54.459228481 [2025-06-11 00:56:54] frigate.ptz.autotrack INFO : Exiting autotracker...2025-06-11 00:56:54.459428081 [2025-06-11 00:56:54] frigate.storage INFO : Exiting storage maintainer...2025-06-11 00:56:54.459971064 [2025-06-11 00:56:54] frigate.watchdog INFO : Exiting watchdog...2025-06-11 00:56:54.459991784 [2025-06-11 00:56:54] frigate.events.cleanup INFO : Exiting event cleanup...2025-06-11 00:56:54.460017042 [2025-06-11 00:56:54] frigate.record.cleanup INFO : Exiting recording cleanup...2025-06-11 00:56:54.460231440 [2025-06-11 00:56:54] frigate.app INFO : Waiting for capture process for doorbell to stop2025-06-11 00:56:54.460845509 [2025-06-11 00:56:54] watchdog.doorbell INFO : Terminating the existing ffmpeg process...2025-06-11 00:56:54.460889012 [2025-06-11 00:56:54] watchdog.doorbell INFO : Waiting for ffmpeg to exit gracefully...2025-06-11 00:56:54.955154575 [2025-06-11 00:56:54] frigate.object_processing INFO : Exiting object processor...2025-06-11 00:56:55.167161959 [2025-06-11 00:56:55] watchdog.doorbell INFO : Terminating the existing ffmpeg process...2025-06-11 00:56:55.167180404 [2025-06-11 00:56:55] watchdog.doorbell INFO : Waiting for ffmpeg to exit gracefully...2025-06-11 00:56:55.236468884 [2025-06-11 00:56:55] frigate.events.maintainer INFO : Exiting event processor...2025-06-11 00:56:55.294615435 [2025-06-11 00:56:55] frigate.app INFO : Waiting for process for doorbell to stop2025-06-11 00:56:58.945911348 [2025-06-11 00:56:58] frigate.video INFO : doorbell: emptying frame queue
June 11, 20251 yr Author 6 hours ago, renk1 said:Fatal Python error: Segmentation faultSeems that your detection is not workingDid you have correct drivers installend in using the nvidia plugin?Are you using the nvidia tensorRT branch in the installation?Did you added "--runtime=nvidia" as extra parameter?Did you correctly passtrought the new GPU?This part doesn't match with documentation:detectors:tensorrt:type: tensorrtdevice: 0model:path: /config/model_cache/tensorrt/8.5.3/yolov7-tiny-416.trtinput_tensor: nchwinput_shape: [1, 3, 416, 416]https://docs.frigate.video/configuration/object_detectors#configuration-parametersYou should change input shape to width: 416height: 416
June 11, 20251 yr 9 hours ago, yayitazale said:Seems that your detection is not workingDid you have correct drivers installend in using the nvidia plugin?Are you using the nvidia tensorRT branch in the installation?Did you added "--runtime=nvidia" as extra parameter?Did you correctly passtrought the new GPU?This part doesn't match with documentation:detectors:tensorrt:type: tensorrtdevice: 0model:path: /config/model_cache/tensorrt/8.5.3/yolov7-tiny-416.trtinput_tensor: nchwinput_shape: [1, 3, 416, 416]https://docs.frigate.video/configuration/object_detectors#configuration-parametersYou should change input shape to width: 416height: 416Correct if I'm wrong but in the logs it mentions some gpu stats for a few seconds before the python segmentation fault, doesn't it mean that is being used? Also, on the frigate gui I could see rtx a4000 utilization on the bottom.I'm using the Nvidia Driver Version: 570.153.02 from the production branch, I'm using the tensorrt branch and already added the --runtime=nvidia , corrected the tensorrt part like you recommended and I'm still getting the same error...I also ran nvidia-smi and it seems fine I thinkEDIT: don't really know if there is anything else I have to do regarding gpu passthrough Edited June 11, 20251 yr by renk1
June 11, 20251 yr While browsing closed issues on github I saw some reports of users pointing that they could only run models up to 320 so I changed to yolov7-320 and it seems to be stable. thanks @yayitazale
June 12, 20251 yr Author 9 hours ago, renk1 said:While browsing closed issues on github I saw some reports of users pointing that they could only run models up to 320 so I changed to yolov7-320 and it seems to be stable. thanks @yayitazaleThanks for the update.
June 25, 20251 yr I tried to search but all I find is people asking the question and no concrete answers. I just switch nvidia gpu from a rtx 2060 to rtx3060 and now it says:The engine plan file is generated on an incompatible device, expecting compute 8.6 got compute 7.5, please rebuildHow do I do this?
June 25, 20251 yr 6 minutes ago, mcmasterp said:I tried to search but all I find is people asking the question and no concrete answers. I just switch nvidia gpu from a rtx 2060 to rtx3060 and now it says:The engine plan file is generated on an incompatible device, expecting compute 8.6 got compute 7.5, please rebuildHow do I do this?All I had to do was post here to then figure it out myself of course lol. for others:you need to delete the folder under config/model_cache/tensorrt/ mine was called 8.5.3
July 5, 2025Jul 5 Hello,maybe a experienced user could give me some hints befor buying stuff, because I do have general questions about hardware to run frigate.My host is a Qnap TS-253d with Unraid v7.1.4.The NAS runs with a J4125 (HD600), 16GB Ram, 1hdd, 2xNvme @ pcie2 card and has got USB2.0 and USB3.2 (Gen1) ports.I do have only 2-4 Cams running at the moment with Motioneye Docker Container. Efficiency is top, because only motion events are recorded, but motion capturing is not very reliable.Shinoby and Zoneminder are working fine as well, motion detection is better, but they need to much recources with 4 cams because of the constant recordings. (2cams max. would be just okay)On the NAS are 3-4 Dockers, 1-2 Linux VM's and samba shares active. Sometimes it runs at its limit, but all works still fine.My Idea was to get a Coral USB Acellerator and attach it with a USB-C to USB3 cable to run with Frigate.(Qnap hompage says it is compatible. https://www.qnap.com/de-de/compatibility/?model=441&category=31 )Would this work without any hardware difficultys and is there something important I have to know?What would be the CPU consumption in comparison to Motioneye?thanks in advance!edit:I found informations about the Odyssey X86 Blue by Seeed Studios from what I think is comparable to the Qnap. And the needed recources seem to be almost the same as with Motioneye.Anything else to know? Edited July 5, 2025Jul 5 by NoRaid99
July 7, 2025Jul 7 Author On 7/5/2025 at 4:18 PM, NoRaid99 said:Hello,maybe a experienced user could give me some hints befor buying stuff, because I do have general questions about hardware to run frigate.My host is a Qnap TS-253d with Unraid v7.1.4.The NAS runs with a J4125 (HD600), 16GB Ram, 1hdd, 2xNvme @ pcie2 card and has got USB2.0 and USB3.2 (Gen1) ports.I do have only 2-4 Cams running at the moment with Motioneye Docker Container. Efficiency is top, because only motion events are recorded, but motion capturing is not very reliable.Shinoby and Zoneminder are working fine as well, motion detection is better, but they need to much recources with 4 cams because of the constant recordings. (2cams max. would be just okay)On the NAS are 3-4 Dockers, 1-2 Linux VM's and samba shares active. Sometimes it runs at its limit, but all works still fine.My Idea was to get a Coral USB Acellerator and attach it with a USB-C to USB3 cable to run with Frigate.(Qnap hompage says it is compatible. https://www.qnap.com/de-de/compatibility/?model=441&category=31 )Would this work without any hardware difficultys and is there something important I have to know?What would be the CPU consumption in comparison to Motioneye?thanks in advance!edit:I found informations about the Odyssey X86 Blue by Seeed Studios from what I think is comparable to the Qnap. And the needed recources seem to be almost the same as with Motioneye.Anything else to know?I have never used Motioneye so I don't know if the CPU usage will be higher or lower. About USB Coral, I have been using one for the last 3 years non-stop with frigate (8 cameras FHD) and it works perfect.
July 7, 2025Jul 7 Thank you, interresting would be to know the hardware you use with the usb accelerator.Anyway, I think my Qnap will do it without reaching its limits.
July 8, 2025Jul 8 Author 12 hours ago, NoRaid99 said:Thank you, interresting would be to know the hardware you use with the usb accelerator.Anyway, I think my Qnap will do it without reaching its limits.I changed the server a few months ago to an Intel Pentium Gold 8505 - 32 GB DDR5 - Integrated Intel i915 Alder Lake-UP3 GT1 GPU.Previously I had an Intel Pentium Gold G5600 - 32 Gb DDR4 - Nvidia Gforce GTX1060 3GB dedicated GPU.With the old server and all cameras running I had about 25-30% CPU usage, and now with the new server I barely see more than 20%. I usually have 13-14% CPU usage with all cameras on during the night, and during the day with 4 cameras I have about 8% CPU usage, with Frigate always being the biggest consumer.Edit: I use the GPU for FFMPEG restreaming. Edited July 8, 2025Jul 8 by yayitazale
July 13, 2025Jul 13 After a few problems at startup, Frigate is working fine on my TS-253d with the USB accelerator.The nastyest problem was the defective USB cable from the Coral device. It just didn't work.Consumption of cpu-resources can be realy low, depends on the capture stream. In general it does not use more recources than MotionEye, but I don't know if this is comparable.I have a few questions:If I run Frigate only on UnRaid and don't use the HA Integration, is it save to cancel the rtsp upstream ports setting when creating the container? (already deleted this and Frigate runs, but maybe this is not good to do?)How can I set a user password, that will be asked for login? I'm logged in as anonymus.On a screenshot on the Frigate homepage I have seen other tabs like "developer tools" in the GUI. Is it possible to activate this somehow?Is there a good free tool to catch all the available rtsp stream links of a camera?Thanks to all being involved somehow into this project! Edited July 13, 2025Jul 13 by NoRaid99
July 13, 2025Jul 13 Hello,since I upgraded frigeate to v15 I have some error log with --> "Too many unprocessed recording segments"The camera recording stops and after a restart the container works for hours sometimes for 1-2 days normally.Have someone a ideas how can I solve that issue? Is there a need to increase the swap?I found that this code in the compose file can solve the issue but how can I add that in unraid:mem_swappiness: 0memswap_limit: 7.5gbdeploy: resources: limits: memory: 7.5gb
July 15, 2025Jul 15 Author On 7/13/2025 at 9:17 AM, NoRaid99 said:After a few problems at startup, Frigate is working fine on my TS-253d with the USB accelerator.The nastyest problem was the defective USB cable from the Coral device. It just didn't work.Consumption of cpu-resources can be realy low, depends on the capture stream. In general it does not use more recources than MotionEye, but I don't know if this is comparable.I have a few questions:If I run Frigate only on UnRaid and don't use the HA Integration, is it save to cancel the rtsp upstream ports setting when creating the container? (already deleted this and Frigate runs, but maybe this is not good to do?)How can I set a user password, that will be asked for login? I'm logged in as anonymus.On a screenshot on the Frigate homepage I have seen other tabs like "developer tools" in the GUI. Is it possible to activate this somehow?Is there a good free tool to catch all the available rtsp stream links of a camera?Thanks to all being involved somehow into this project!If you don't plan to use this rtsp streams anywhere else, then I think there is no problem to delete them.https://docs.frigate.video/configuration/authentication#onboardingWith a Admin userYou can try https://www.advanced-ip-scanner.com/es/
July 15, 2025Jul 15 Author On 7/13/2025 at 5:32 PM, snowy00 said:Hello,since I upgraded frigeate to v15 I have some error log with --> "Too many unprocessed recording segments"The camera recording stops and after a restart the container works for hours sometimes for 1-2 days normally.Have someone a ideas how can I solve that issue? Is there a need to increase the swap?I found that this code in the compose file can solve the issue but how can I add that in unraid:mem_swappiness: 0memswap_limit: 7.5gbdeploy:resources:limits:memory: 7.5gbEdit the container, click on advanced view (top right) and then you have to edit extra parameters:This is my config for 6 full hd cams:--shm-size=562m --mount type=tmpfs,target=/tmp/cache,tmpfs-size=1000000000 --restart unless-stoppedFor SHM size calculation: https://docs.frigate.video/frigate/installation#calculating-required-shm-sizeFor tmpfs there is no guide, I'm currently using 1GB with no issues.
July 15, 2025Jul 15 24 minutes ago, yayitazale said:If you don't plan to use this rtsp streams anywhere else, then I think there is no problem to delete them.https://docs.frigate.video/configuration/authentication#onboardingWith a Admin userYou can try https://www.advanced-ip-scanner.com/es/Thanks a lot for your helpfull answers, that is very kind.about point 2:I found the docs for auth and I can login with credentials on port 8971. But the problem is that port 5000 is allways open to reach GUI. I created a new container without the port 5000 setting variale, but it is still set and open.Is there a way to block this port? Otherwise it doesn't seem to make much sense for me.(Perhaps it makes a difference that I have the container connectet to "eth" and gaved him his own IP.)point3:Logged in to admin account there is only a additional notification setting.Is the "developer tools" on the screenshot from older builds and not active anymore?edit: Finally I realised that this screenshot is from HA gui. 🧐one more question:I do isolate iot devices and block them from connecting to www, also frigate.On starup there is "frigate.util.downloader" trying to connect to "huggingface.co". What is that for and is this connection needed if no cloud services etc. are activated?have a nice day! Edited July 15, 2025Jul 15 by NoRaid99
July 15, 2025Jul 15 Author 4 minutes ago, NoRaid99 said:Thanks a lot for your helpfull answers, that is very kind.about point 2:I found the docs for auth and I can login with credentials on port 8971. But the problem is that port 5000 is allways open to reach GUI. I created a new container without the port 5000 setting variale, but it is still set and open.Is there a way to block this port? Otherwise it doesn't seem to make much sense for me.(Perhaps it makes a difference that I have the container connectet to "eth" and gaved him his own IP.)point3:Logged in to admin account there is only a additional notification setting.Is the "developer tools" on the screenshot from older builds and not active anymore?edit: Finally I realised that this screenshot is from HA gui. 🧐one more question:I do isolate iot devices and block them from connecting to www, also frigate.On starup there is "frigate.util.downloader" trying to connect to "huggingface.co". What is that for and is this connection needed if no cloud services etc. are activated?have a nice day!Port 5000 is a legacy port and will be only used for API in the future. I'm not part of the developers team of the project but I assume that the connection to "huggingface.co" is done to donwload semantic search models https://docs.frigate.video/configuration/semantic_search. Edited July 15, 2025Jul 15 by yayitazale
July 15, 2025Jul 15 5 hours ago, yayitazale said:Port 5000 is a legacy port and will be only used for API in the future.Does this depend to the developers from frigate or unraid?I have the strange feeling that this future should be now. :)5 hours ago, yayitazale said:I'm not part of the developers team of the project but I assume that the connection to "huggingface.co" is done to donwload semantic search models https://docs.frigate.video/configuration/semantic_search.At the moment semantic search set to "small model size" is working in a new created container never started with internet connection.I will keep the setting like this and observe if there are be problems in future.
July 15, 2025Jul 15 I'm having an issue where I click "install" on the docker image, fill out all of the info and then when it's pulling down the images it seems to hang on one part and never completes. Has anyone else run into something like that?
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.