[Support]: ffmpeg crashing causing OOM kills #14571
Replies: 5 comments 4 replies
-
|
Forgot to add that the container usually uses between 1.5-2.5GB of memory during normal running |
Beta Was this translation helpful? Give feedback.
-
|
You can try updating to a newer version of ffmpeg, some users have said this fixed the issue for them. The next version of frigate will use a newer version by default. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the response! I've tried to follow the documentation and have done the following to upgrade ffmpeg: Extracted this to the host: Then confirmed this is the folder that contains the /bin folder: And then mapped that to the container at: extraVolumes:
extraVolumeMounts:
Unfortunately the container fails with these logs: |
Beta Was this translation helpful? Give feedback.
-
|
you have to use a statically built ffmpeg binary |
Beta Was this translation helpful? Give feedback.
-
|
I have had a very similar issue to this about 6 months ago: #10814 Also using Hikvision cameras, and resizing the stream manually using go2rtc. Unfortunately I've had to keep ffmpeg at 5.0 due to other issues; will have a look through my logs at whether this issue is still happening. I moved some of my Wifi cameras to ethernet which may have been the root cause of the slightly messy camera dropouts that go2rtc/ffmpeg were getting hung up on. |
Beta Was this translation helpful? Give feedback.
-
|
im kind of a noob but had a similar oom issue that i think i finally solved just last night. it was actually two fixes. 1) i'm running frigate on proxmox in a portainer container and noticed the container reports 4.3gb of memory although i only have 4gb allocated to the lxc. in portainer i changed the resource mapping to be around 3.8gb. 2) my reolink cameras were sending the detect stream in h.264 and the main recording stream in h.265. there is no way to change this on camera due to the higher resolution they say. In the frigate config, I had to enable different hardware accel for each stream (not globally) and specify hevc or h.264 respectively. since doing that i no longer have any ffmpeg errors and no more oom crashes. i'm not using go2rtc at all (at least i dont think so). not sure if this is helpful but I've seen the oom issue in a few posts now. |
Beta Was this translation helpful? Give feedback.
-
|
I'm curious if the people seeing this can check some things. For me, the actual used memory by frigate (and associated programs) isn't that high, a couple of gigs, however the cache memory, ie the memory that the kernel has available to free if it needs it, but is holding on to it for cleanup later, grows very high. There are some metrics that help us identify if this is an issue: These are counters, so you need to check periodically, ideally after you have seen the oomkiller trigger. What do the numbers mean? well, kswapd is the kernel process that takes pages that are in cache or inactive, ie available to be converted to free memory, and the two "hit quickly" counters are the times when kswapd runs, and 100ms later the amount of free pages is lower than the configured watermark. Basically an indication that kswapd is struggling to keep up with demand. If kswapd completely fails to keep up, and the free pages drops below "min_free_kbytes", then memory requests trigger "direct" page scans, ie pages are attempted to be freed synchronously with the calling application. This can then trigger low memory conditions and the oomkiller. This is all a long winded way of saying, from what I see this isn't a memory leak, but rather the linux kernel not dealing well with a constant stream of allocations. I have changed /proc/sys/vm/watermark_scale_factor to 1000, which forces kswapd to act earlier and more aggressively and the result is this: There may be other ways to achieve this, and this is probably overly aggressive, but for now this seems to have resolved the issue for me. Given this system is dedicated to running Frigate, I'm not worried about the settings being overly aggressive. For a more multi-purpose system, I'd probably only set scale factor to 500, and maybe reduce swappiness to 10. This can be achieved by adding: to /etc/sysctl.conf and rebooting. |
Beta Was this translation helpful? Give feedback.
-
|
Some really interesting analysis right there! What kernel version are you using out of interest? Can you tell what is occupying the cache memory? I think you might have a different issue to many of us though. Many cases of OOMkiller being invoked often seem to be ffmpeg consuming gigabytes of RAM per second, rather than over hours. This seems likely to be some odd behaviour in ffmpeg when a connection is flakey (it's often seen on wireless cameras) and it spams the output with vast amounts of data. Your issue looks like it grows over 3-4 hours? |
Beta Was this translation helpful? Give feedback.
-
|
The time between oomkiller running would vary, and wasn't always triggered by ffmpeg: You can see anything from 2 minutes to multiple hours. I strongly suspect it is related to the quality of the stream. I had one camera on wifi which would drop, and I think it contributed to the memory usage. Kernel version: This was a clean, vanilla install of Debian just to run Frigate, outside of installing dependencies, I made no changes to the OS, well, before this one. The oomkiller message itself contains the information about where the memory is being used (which I now notice no-one has actually posted! Can people share examples of their syslog showing the oomkiller message please?): There is a lot here, but things to note: So in my case at least, I'm reasonably confident that the solution here is a decent one. It's been a week now since I have had an oomkiller event. Whether this is the same as described in the parent post, you are right, I don't know. Getting the oomkiller log from syslog would help tho. EDIT: I re-read your comment about wifi cameras. I definitely agree with that that the issue appears related to flakey connections. I have a single wifi camera only, which may explain the different rate of occurrence (and I've since moved it to wired). |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Describe the problem you are having
Hi there,
I've been struggling with this issue for quite sometime. Originally I had been running Frigate as a container on Unraid, but ran into some strange issues with my server dying overnight. After finally getting the logs persisted I found that the host was being killed by OOM errors and was not able to enforce the memory limit I had set via docker args (--memory=8G). After some additional digging I believe that the cause of this was ffmpeg crashing and requesting a huge amount of memory (much more than th 32GB of the host).
My leading theory is that my cameras have inconsistent streams that constantly force ffmpeg to either restart gracefully or crash each time. Before properly understanding Frigate, I had just bought what I thought to be decent used IP cameras. It was only later that I discovered that most of my cameras don't have high enough sub stream resolutions to be useful for detection. I somehow got around this by downscaling the main stream as a substream to use for detect.
I have now moved Frigate onto kubernetes using the helm template with argocd, all running on proxmox VMs. Now I don't have the issue of the OOM errors killing the entire host, but instead just restarting the container. I had left it running overnight and saw that the container was killed 8 times within the space of about 7 hours. I've tried most of the options I could find to try and make it more stable but don't really have the knowledge to do so with confidence.
Any and all advice would be greatly appreciated along with some mercy when it comes to my config :)
Version
0.14.1-f4f3cfa
What browser(s) are you using?
No response
Frigate config file
Relevant Frigate log output
Relevant go2rtc log output
FFprobe output from your camera
For some reason my ffprobes timeout from within the container and via my localFrigate stats
{ "cameras": { "BackYardCam": { "audio_dBFS": 0.0, "audio_rms": 0.0, "camera_fps": 5.1, "capture_pid": 481, "detection_enabled": true, "detection_fps": 0.0, "ffmpeg_pid": 653, "pid": 457, "process_fps": 5.1, "skipped_fps": 0.0 }, "GateCam": { "audio_dBFS": 0.0, "audio_rms": 0.0, "camera_fps": 5.0, "capture_pid": 479, "detection_enabled": true, "detection_fps": 0.0, "ffmpeg_pid": 2885, "pid": 453, "process_fps": 5.0, "skipped_fps": 0.0 }, "TestCam": { "audio_dBFS": 0.0, "audio_rms": 0.0, "camera_fps": 6.0, "capture_pid": 493, "detection_enabled": true, "detection_fps": 0.0, "ffmpeg_pid": 505, "pid": 464, "process_fps": 6.0, "skipped_fps": 0.0 }, "YardCam": { "audio_dBFS": 0.0, "audio_rms": 0.0, "camera_fps": 6.1, "capture_pid": 473, "detection_enabled": true, "detection_fps": 10.3, "ffmpeg_pid": 650, "pid": 446, "process_fps": 6.1, "skipped_fps": 0.0 } }, "cpu_usages": { "1": { "cmdline": "/package/admin/s6/command/s6-svscan -d4 -- /run/service", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "103": { "cmdline": "s6-log -b -- T 1 n0 s10000000 T /dev/shm/logs/go2rtc", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "104": { "cmdline": "s6-log -b -- T 1 n0 s10000000 T /dev/shm/logs/frigate", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "105": { "cmdline": "s6-log -b -- T 1 n0 s10000000 T /dev/shm/logs/certsync", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "113": { "cmdline": "/usr/local/go2rtc/bin/go2rtc -config=/dev/shm/go2rtc.yaml", "cpu": "13.5", "cpu_average": "12", "mem": "0.5" }, "127": { "cmdline": "bash ./run.user go2rtc-healthcheck", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "143": { "cmdline": "python3 -u -m frigate", "cpu": "4.6", "cpu_average": "4", "mem": "7.6" }, "149": { "cmdline": "nginx: master process nginx", "cpu": "0.0", "cpu_average": "0", "mem": "0.2" }, "177": { "cmdline": "nginx: worker process", "cpu": "0.0", "cpu_average": "0", "mem": "0.1" }, "178": { "cmdline": "nginx: worker process", "cpu": "0.0", "cpu_average": "0", "mem": "0.1" }, "179": { "cmdline": "nginx: worker process", "cpu": "0.0", "cpu_average": "0", "mem": "0.1" }, "180": { "cmdline": "nginx: worker process", "cpu": "0.0", "cpu_average": "0", "mem": "0.1" }, "182": { "cmdline": "nginx: cache manager process", "cpu": "0.0", "cpu_average": "0", "mem": "0.1" }, "23": { "cmdline": "s6-supervise s6-linux-init-shutdownd", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "25": { "cmdline": "/package/admin/s6-linux-init/command/s6-linux-init-shutdownd -c /run/s6/basedir -g 3000 -C -B", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "2885": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/GateCam_sub -r 5 -vf fps=5,scale_cuda=w=1920:h=1080:format=nv12,hwdownload,format=nv12,format=yuv420p -threads 2 -f rawvideo -pix_fmt yuv420p pipe:", "cpu": "5.2", "cpu_average": "4", "mem": "2.4" }, "2887": { "cmdline": "ffmpeg -hide_banner -v error -hwaccel cuda -hwaccel_output_format cuda -fflags nobuffer -flags low_delay -timeout 5000000 -user_agent go2rtc/ffmpeg -rtsp_flags prefer_tcp -i rtsp://127.0.0.1:8554/GateCam?video -r 5 -c:v h264_nvenc -g 50 -bf 0 -profile:v high -level:v auto -preset:v p2 -tune:v ll -an -vf scale_cuda=1280:720 -user_agent ffmpeg/go2rtc -rtsp_transport tcp -f rtsp rtsp://127.0.0.1:8554/307305e8d4b969067924adeef130238e", "cpu": "2.2", "cpu_average": "2", "mem": "2.5" }, "3097": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/GateCam -f segment -segment_time 10 -segment_format mp4 -reset_timestamps 1 -strftime 1 -c copy -an /tmp/cache/GateCam@%Y%m%d%H%M%S%z.mp4", "cpu": "0.9", "cpu_average": "0", "mem": "0.3" }, "314": { "cmdline": "bash ./run.user certsync", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "3166": { "cmdline": "/bin/sh", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "32": { "cmdline": "s6-supervise nginx", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "33": { "cmdline": "s6-supervise certsync", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "34": { "cmdline": "s6-supervise frigate", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "35": { "cmdline": "s6-supervise go2rtc-healthcheck", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "357": { "cmdline": "frigate.logger ", "cpu": "0.0", "cpu_average": "0", "mem": "2.4" }, "359": { "cmdline": "frigate.recording_manager", "cpu": "2.7", "cpu_average": "3", "mem": "3.0" }, "36": { "cmdline": "s6-supervise go2rtc-log", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "360": { "cmdline": "frigate.review_segment_manager", "cpu": "0.7", "cpu_average": "0", "mem": "2.9" }, "3695": { "cmdline": "sleep 60", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "37": { "cmdline": "s6-supervise frigate-log", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "3724": { "cmdline": "sleep 30s", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "3725": { "cmdline": "ffprobe rtsp://*:*@192.168.1.79:554/streaming/channels/101?tcp", "cpu": "0.1", "cpu_average": "0", "mem": "0.2" }, "38": { "cmdline": "s6-supervise go2rtc", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "387": { "cmdline": "/usr/bin/python3 -c from multiprocessing.resource_tracker import main;main(65)", "cpu": "0.3", "cpu_average": "0", "mem": "0.1" }, "388": { "cmdline": "frigate.detector.tensorrt", "cpu": "44.0", "cpu_average": "42", "mem": "7.5" }, "39": { "cmdline": "s6-supervise nginx-log", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "390": { "cmdline": "frigate.output ", "cpu": "2.9", "cpu_average": "2", "mem": "2.8" }, "395": { "cmdline": "ffmpeg -threads 1 -f rawvideo -pix_fmt yuv420p -video_size 1280x720 -i pipe: -threads 1 -f mpegts -s 1280x720 -codec:v mpeg1video -q 8 -bf 0 pipe:", "cpu": "0.0", "cpu_average": "0", "mem": "0.2" }, "40": { "cmdline": "s6-supervise s6rc-fdholder", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "406": { "cmdline": "ffmpeg -threads 1 -f rawvideo -pix_fmt yuv420p -video_size 1920x1080 -i pipe: -threads 1 -f mpegts -s 1280x720 -codec:v mpeg1video -q 8 -bf 0 pipe:", "cpu": "0.0", "cpu_average": "0", "mem": "0.2" }, "41": { "cmdline": "s6-supervise s6rc-oneshot-runner", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "411": { "cmdline": "ffmpeg -threads 1 -f rawvideo -pix_fmt yuv420p -video_size 1920x1080 -i pipe: -threads 1 -f mpegts -s 1280x720 -codec:v mpeg1video -q 8 -bf 0 pipe:", "cpu": "0.0", "cpu_average": "0", "mem": "0.2" }, "417": { "cmdline": "ffmpeg -threads 1 -f rawvideo -pix_fmt yuv420p -video_size 1280x720 -i pipe: -threads 1 -f mpegts -s 1280x720 -codec:v mpeg1video -q 8 -bf 0 pipe:", "cpu": "0.0", "cpu_average": "0", "mem": "0.2" }, "42": { "cmdline": "s6-supervise certsync-log", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "423": { "cmdline": "ffmpeg -threads 1 -f rawvideo -pix_fmt yuv420p -video_size 1280x720 -i pipe: -threads 1 -f mpegts -s 1280x720 -codec:v mpeg1video -q 8 -bf 0 pipe:", "cpu": "0.0", "cpu_average": "0", "mem": "0.2" }, "446": { "cmdline": "frigate.process:YardCam", "cpu": "9.5", "cpu_average": "6", "mem": "3.0" }, "453": { "cmdline": "frigate.process:GateCam", "cpu": "1.4", "cpu_average": "2", "mem": "3.5" }, "457": { "cmdline": "frigate.process:BackYardCam", "cpu": "1.5", "cpu_average": "1", "mem": "2.9" }, "464": { "cmdline": "frigate.process:TestCam", "cpu": "1.7", "cpu_average": "2", "mem": "2.9" }, "473": { "cmdline": "frigate.capture:YardCam", "cpu": "1.8", "cpu_average": "1", "mem": "2.8" }, "479": { "cmdline": "frigate.capture:GateCam", "cpu": "2.9", "cpu_average": "2", "mem": "2.8" }, "481": { "cmdline": "frigate.capture:BackYardCam", "cpu": "2.9", "cpu_average": "2", "mem": "2.8" }, "493": { "cmdline": "frigate.capture:TestCam", "cpu": "1.8", "cpu_average": "1", "mem": "2.8" }, "496": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/YardCam -f segment -segment_time 10 -segment_format mp4 -reset_timestamps 1 -strftime 1 -c copy -an /tmp/cache/YardCam@%Y%m%d%H%M%S%z.mp4", "cpu": "0.8", "cpu_average": "0", "mem": "0.3" }, "505": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/TestCam_sub -r 6 -vf fps=6,scale_cuda=w=1280:h=720:format=nv12,hwdownload,format=nv12,format=yuv420p -threads 2 -f rawvideo -pix_fmt yuv420p pipe:", "cpu": "3.2", "cpu_average": "3", "mem": "3.8" }, "514": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/BackYardCam -f segment -segment_time 10 -segment_format mp4 -reset_timestamps 1 -strftime 1 -c copy -an /tmp/cache/BackYardCam@%Y%m%d%H%M%S%z.mp4", "cpu": "0.9", "cpu_average": "0", "mem": "0.3" }, "517": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/TestCam -f segment -segment_time 10 -segment_format mp4 -reset_timestamps 1 -strftime 1 -c copy -an /tmp/cache/TestCam@%Y%m%d%H%M%S%z.mp4", "cpu": "0.7", "cpu_average": "0", "mem": "0.3" }, "53": { "cmdline": "/package/admin/s6/command/s6-ipcserverd -1 -- /package/admin/s6/command/s6-ipcserver-access -v0 -E -l0 -i data/rules -- /package/admin/s6/command/s6-sudod -t 30000 -- /package/admin/s6-rc/command/s6-rc-oneshot-run -l ../.. --", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "54": { "cmdline": "/package/admin/s6-2.11.3.2/command/s6-fdholderd -1 -i data/rules", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "639": { "cmdline": "/usr/bin/python3 -c from multiprocessing.resource_tracker import main;main(54)", "cpu": "0.0", "cpu_average": "0", "mem": "0.1" }, "650": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/YardCam_sub -r 6 -vf fps=6,scale_cuda=w=1280:h=720:format=nv12,hwdownload,format=nv12,format=yuv420p -threads 2 -f rawvideo -pix_fmt yuv420p pipe:", "cpu": "2.3", "cpu_average": "2", "mem": "2.6" }, "652": { "cmdline": "ffmpeg -hide_banner -v error -hwaccel cuda -hwaccel_output_format cuda -fflags nobuffer -flags low_delay -timeout 5000000 -user_agent go2rtc/ffmpeg -rtsp_flags prefer_tcp -i rtsp://127.0.0.1:8554/YardCam?video -r 5 -c:v h264_nvenc -g 50 -bf 0 -profile:v high -level:v auto -preset:v p2 -tune:v ll -an -vf scale_cuda=1920:1080 -user_agent ffmpeg/go2rtc -rtsp_transport tcp -f rtsp rtsp://127.0.0.1:8554/de293f899117f8fd374c19b1bf00894f", "cpu": "1.9", "cpu_average": "1", "mem": "2.6" }, "653": { "cmdline": "ffmpeg -hide_banner -loglevel warning -hwaccel cuda -hwaccel_output_format cuda -user_agent FFmpeg Frigate/0.14.1-f4f3cfa -rtsp_transport tcp -timeout 5000000 -i rtsp://127.0.0.1:8554/BackYardCam_sub -r 5 -vf fps=5,scale_cuda=w=1920:h=1080:format=nv12,hwdownload,format=nv12,format=yuv420p -threads 2 -f rawvideo -pix_fmt yuv420p pipe:", "cpu": "5.2", "cpu_average": "5", "mem": "2.6" }, "657": { "cmdline": "ffmpeg -hide_banner -v error -hwaccel cuda -hwaccel_output_format cuda -fflags nobuffer -flags low_delay -timeout 5000000 -user_agent go2rtc/ffmpeg -rtsp_flags prefer_tcp -i rtsp://127.0.0.1:8554/BackYardCam?video -r 5 -c:v h264_nvenc -g 50 -bf 0 -profile:v high -level:v auto -preset:v p2 -tune:v ll -an -vf scale_cuda=1920:1080 -user_agent ffmpeg/go2rtc -rtsp_transport tcp -f rtsp rtsp://127.0.0.1:8554/c4dad4b38df0f106cf1d3b53cd38d19b", "cpu": "2.0", "cpu_average": "2", "mem": "2.7" }, "658": { "cmdline": "ffmpeg -hide_banner -v error -hwaccel cuda -hwaccel_output_format cuda -fflags nobuffer -flags low_delay -timeout 5000000 -user_agent go2rtc/ffmpeg -rtsp_flags prefer_tcp -i rtsp://127.0.0.1:8554/GateCam?video -r 5 -c:v h264_nvenc -g 50 -bf 0 -profile:v high -level:v auto -preset:v p2 -tune:v ll -an -vf scale_cuda=1280:720 -user_agent ffmpeg/go2rtc -rtsp_transport tcp -f rtsp rtsp://127.0.0.1:8554/307305e8d4b969067924adeef130238e", "cpu": "1.9", "cpu_average": "1", "mem": "2.5" }, "99": { "cmdline": "s6-log -b -- T 1 n0 s10000000 T /dev/shm/logs/nginx", "cpu": "0.0", "cpu_average": "0", "mem": "0.0" }, "frigate.full_system": { "cpu": "37.1", "mem": "52.7" } }, "detection_fps": 10.3, "detectors": { "tensorrt": { "detection_start": 1729839163.353684, "inference_speed": 47.82, "pid": 388 } }, "gpu_usages": { "NVIDIA GeForce GTX 1070": { "dec": "21.0%", "enc": "1.0%", "gpu": "18.0%", "mem": "31.31%" } }, "processes": { "go2rtc": { "pid": 113 }, "logger": { "pid": 357 }, "recording": { "pid": 359 }, "review_segment": { "pid": 360 } }, "service": { "last_updated": 1729839163, "latest_version": "0.14.1", "storage": { "/dev/shm": { "free": 932.4, "mount_type": "tmpfs", "total": 1024.0, "used": 91.6 }, "/media/frigate/clips": { "free": 1630214.0, "mount_type": "nfs4", "total": 2860202.0, "used": 1229988.0 }, "/media/frigate/recordings": { "free": 1630214.0, "mount_type": "nfs4", "total": 2860202.0, "used": 1229988.0 }, "/tmp/cache": { "free": 930.3, "mount_type": "tmpfs", "total": 1024.0, "used": 93.7 } }, "temperatures": {}, "uptime": 914, "version": "0.14.1-f4f3cfa" } }Install method
Proxmox via Docker
docker-compose file or Docker CLI command
Object Detector
TensorRT
Network connection
Wired
Camera make and model
Hikvision and TP-Link
Screenshots of the Frigate UI's System metrics pages
Any other information that may be helpful
I know that my inference speed is quite slow with the large model, but as I've just moved the system over, I'm running with my test GPU. The normal GPU I use is a 1660 super with inference times just under 30ms with the largest model
Beta Was this translation helpful? Give feedback.
All reactions