[Bug]: In mobile view event playback jumps to the wrong point in time #20970
-
Checklist
Describe the problem you are havingThis is similar to this discussion, however for me it's only happening in the mobile view. When I clcik on the even it jumps to the wrong place in the timeline. It jumps to some motion recorded after the actual event (in my case usually 15-20 min later). In desktop view everything is fine. Strangely, it seems to be directly related to the mobile view and not to Android/phone/browser/etc. If I emulate a mobile browser on desktop Chrome it does the same thing. Likewise, if I emulate a desktop browser on Android Chrome the issue doesn't happen. Here is a comparison of desktop and mobile view for the same event, where a van comes up the driveway. In desktop view it works fine. In mobile view it skips over that event and goes to the next recorded motion. The event is at 11:33AM, but in mobile view it skips to 12:01PM. This is recorded on the desktop Chrome, just with emulated mobile view. Same happens on Chrome on Android. Desktop view: Mobile view: The recordings are done on the same machine, same browser. The second just uses the emulated mobile view in Chrome webtools. The same issue happens in Chrome in Android. And similarily, if I force a desktop view in Chrome Android, there is no issue. Unfortunately this is quite a big problem for me since other family members exclusively use Frigate on phones, and I'm already hearing complaints like "why can't we just use the Reolink app like before"... Steps to reproduce
Version0.16.2-4d58206 In which browser(s) are you experiencing the issue with?Google Chrome Version 142.0.7444.163 (Official Build) (64-bit), Google Chrome Android 142.0.7444.159 Frigate config filemqtt:
enabled: true
host: 192.168.2.10
user: '{FRIGATE_MQTT_USER}'
password: '{FRIGATE_MQTT_PASSWORD}'
auth:
session_length: 604800 # 7 days
refresh_time: 345600 # 4 days
model:
path: plus://REDACTED_PLUS_ID
detectors:
coral1:
type: edgetpu
device: pci:0
coral2:
type: edgetpu
device: pci:1
objects:
track:
- person
- face
- baby
- car
- license_plate
- bicycle
- motorcycle
- dog
- cat
- bird
- fox
- mustelid
filters:
person:
min_score: 0.56
threshold: 0.68 #0.70
car:
threshold: 0.90
cat:
min_score: 0.4 #0.45
threshold: 0.6 #0.65
license_plate:
min_score: 0.4
review:
alerts:
labels:
- person
- car
- bicycle
- motorcycle
- cat
- dog
- fox
- mustelid
record:
enabled: true
sync_recordings: false
retain:
days: 14
mode: motion
alerts:
pre_capture: 20
post_capture: 20
retain:
days: 30
mode: motion
detections:
pre_capture: 10
post_capture: 10
retain:
days: 30
mode: motion
snapshots:
enabled: true
clean_copy: true
timestamp: false
retain:
default: 30
go2rtc:
streams:
front_yard_camera:
- ffmpeg:http://192.168.20.101/flv?port=1935&app=bcs&stream=channel0_main.bcs&user={FRIGATE_RTSP_USER}&password={FRIGATE_RTSP_PASSWORD}#video=copy#audio=copy#hardware=vaapi
- rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.20.101:554/Preview_01_main
webrtc:
candidates:
- 192.168.20.2:8555
- stun:8555
ffmpeg:
hwaccel_args: preset-intel-qsv-h264
input_args: preset-rtsp-restream
semantic_search:
enabled: false
reindex: false
model_size: large
genai:
enabled: true
provider: ollama
base_url: http://192.168.2.50:11434
model: lukaspetrik/gemma3-tools:12b
cameras:
front_yard_camera:
enabled: true
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/front_yard_camera
roles:
- record
- detect
onvif:
host: 192.168.20.101
port: 8000
user: '{FRIGATE_ONVIF_USER}'
password: '{FRIGATE_ONVIF_PASSWORD}'
live:
streams:
front_yard_camera: front_yard_camera
detect:
enabled: true
width: 1440
height: 1080
motion:
mask:
0,0.212,0.222,0.076,0.259,0.096,0.442,0.086,0.451,0.077,0.563,0.076,0.57,0.087,1,0.085,1,0,0.003,0.001
threshold: 37
contour_area: 12
improve_contrast: true
lightning_threshold: 0.9
zones: {}
objects:
filters:
person:
mask: 0.582,0.058,0.357,0.06,0.352,0,0.584,0
car:
mask: 0.738,0.075,0.742,0,0,0,0,0.083
motorcycle:
mask: 0.567,0.061,0.568,0,0.222,0.001,0.223,0.07
genai:
enabled: true
tls:
enabled: false
logger:
default: warning
version: 0.16-0
detect:
enabled: true
fps: 5
max_disappeared: 50
face_recognition:
enabled: true
min_area: 600
model_size: large
save_attempts: 200
lpr:
enabled: true
model_size: small
recognition_threshold: 0.6
enhancement: 0
debug_save_plates: false
device: GPU
known_plates:
Kia:
- REDACTED_PLATES_REGEX
Hyundai:
- REDACTED_PLATES_REGEX
classification:
bird:
enabled: falsedocker-compose file or Docker CLI commandservices:
frigate:
container_name: frigate
privileged: true
restart: unless-stopped
stop_grace_period: 30s
image: ghcr.io/blakeblackshear/frigate:stable-tensorrt
shm_size: "128mb"
devices:
- /dev/apex_0:/dev/apex_0
- /dev/apex_1:/dev/apex_1
- /dev/dri/renderD128:/dev/dri/renderD128
- /dev/dri/card0:/dev/dri/card0
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: 1 # number of GPUs
capabilities: [gpu]
cap_add:
- CAP_PERFMON
volumes:
- /etc/localtime:/etc/localtime:ro
- ./config:/config
- /media/frigate:/media/frigate
- type: tmpfs
target: /tmp/cache
tmpfs:
size: 1000000000
ports:
- "8971:8971"
- "5000:5000"
- "8554:8554"
- "8555:8555/tcp"
- "8555:8555/udp"
environment:
FRIGATE_RTSP_USER: REDACTED_USER
FRIGATE_RTSP_PASSWORD: REDACTED_PASSWORD
FRIGATE_MQTT_USER: REDACTED_USER
FRIGATE_MQTT_PASSWORD: REDACTED_PASSWORD
FRIGATE_ONVIF_USER: REDACTED_USER
FRIGATE_ONVIF_PASSWORD: REDACTED_PASSWORD
secrets:
- PLUS_API_KEY
secrets:
PLUS_API_KEY:
file: ./frigate_plus_key.txtRelevant Frigate log output2025-11-19 10:36:50.152479227 [INFO] Preparing Frigate...
2025-11-19 10:36:50.666645590 [INFO] Starting Frigate...
2025-11-19 10:36:53.530372239 [2025-11-19 10:36:53] frigate.util.config INFO : Checking if frigate config needs migration...
2025-11-19 10:36:53.548111777 [2025-11-19 10:36:53] frigate.util.config INFO : frigate config does not need migration...
2025-11-19 10:36:53.586921312 [2025-11-19 10:36:53] frigate.config.config WARNING : front_yard_camera is configured to track ['baby', 'mustelid'] objects, which are not supported by the current model.
2025-11-19 10:36:53.650866625 INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
2025-11-19 10:36:53.650866625 [LOGGING] Last message repeated 1 times
2025-11-19 10:36:54.242862948 �[0;93m2025-11-19 10:36:54.242719212 [W:onnxruntime:, transformer_memcpy.cc:74 ApplyImpl] 2 Memcpy nodes are added to the graph main_graph for CUDAExecutionProvider. It might have negative impact on performance (including unable to run CUDA graph). Set session_options.log_severity_level=1 to see the detail logs before this message.�[m
2025-11-19 10:36:54.269602129 loading data from : /config/model_cache/facedet/landmarkdet.yaml
2025-11-19 18:28:14.628758510 [2025-11-19 18:28:14] frigate.genai.ollama WARNING : Ollama returned an error: timed out
2025-11-19 18:28:14.628758510 [LOGGING] Last message repeated 7 timesRelevant go2rtc log output2025-11-19 10:36:50.666517196 [INFO] Preparing new go2rtc config...
2025-11-19 10:36:50.947494951 [INFO] Starting go2rtc...
2025-11-19 10:36:51.051226114 10:36:51.051 INF go2rtc platform=linux/amd64 revision=fa580c5 version=1.9.9
2025-11-19 10:36:51.051228313 10:36:51.051 INF config path=/dev/shm/go2rtc.yaml
2025-11-19 10:36:51.051440925 10:36:51.051 INF [rtsp] listen addr=:8554
2025-11-19 10:36:51.051488810 10:36:51.051 INF [api] listen addr=:1984
2025-11-19 10:36:51.051572234 10:36:51.051 INF [webrtc] listen addr=:8555
2025-11-19 10:37:00.049497662 [INFO] Starting go2rtc healthcheck service...Operating systemProxmox Install methodDocker Compose Network connectionWired Camera make and modelReolink E1 Outdoor Screenshots of the Frigate UI's System metrics pages
Any other information that may be helpfulNo response |
Beta Was this translation helpful? Give feedback.
Replies: 9 comments 10 replies
-
|
I've never seen any issue like this, so it may still be something specific. We have also had a lot of adjustments and fixes in 0.17 for HLS playback, seeking, etc. so this may be fixed in the next version already |
Beta Was this translation helpful? Give feedback.
-
|
I don't know if it's specific or not, but for me it behaves like this for every event recording, it's not just some sporadic ones. Fingers crossed it's fixed in the next release. |
Beta Was this translation helpful? Give feedback.
-
|
that's more weird then because I don't see this issue on the Frigate demo site (which runs the current version) or my personal instance which runs the latest commit |
Beta Was this translation helpful? Give feedback.
-
|
It happens on v0.17 as well. It jumps to the first event in the next "hour file". I will try emulated "mobile view" on desktop firefox next. So far, i just have the bug on Android/Chrome. |
Beta Was this translation helpful? Give feedback.
-
|
Good observation, I didn't notice that before. But you are right, it jumps to the first clip in the next hour. It's always around xx:00, xx:01, xx:02... Although after trying this now on dozens of events, a selected few of them do jump to the correct time. But those are very rare. I haven't figured out the pattern of those yet. Maybe if there is no recording on the begging of the next hour, but I'm not sure. I'll try to figure out if there is a pattern. |
Beta Was this translation helpful? Give feedback.
-
|
We just pushed some changes to the dev branch. Could you test the latest build and see if it resolves your issue? If not, @NickM-27 has a custom build (based on |
Beta Was this translation helpful? Give feedback.
-
|
dev/0.17.0-8e83460 didn't help. I will switch to @NickM-27 build. |
Beta Was this translation helpful? Give feedback.
-
|
My build will be identical to dev at this point. I think we'll need a video or something to see what you're experiencing |
Beta Was this translation helpful? Give feedback.
-
|
This just keeps getting more weird... I did try this initially in an incodnito window on desktop, and I had the problem in the emulated mobile view. However, after clearing everything I could possibly find in chrome devtools: cookies, local storage, session storage, pwa data, cache (not in incognito).... the problem went away on desktop, even in emulated mobile view. Even more weird, after that it went away in Android Chrome as well, even though I didn't clear anything there. And even more weird, it doesn't go away in the PWA app. Works in Chrome on Android, but not in the PWA app. I've tried clearing data and cache of the app, didn't help. I tried uninstalling and reinstalling the app, didn't help. When I switch to forced desktop mode in Android, and open the PWA, it's fine. When I switch back to normal/mobile view, I still have the issue. I even tried installing the app in desktop view, then switching to mobile view... didn't help. Next thing I'll try is to hook up the remote adb debugger and try clearing everything manually from Chrome, like I did on desktop. I'll post if it helped. I know it sounds weird, and I can't find any logical explanation behind it, but that is what happened. Current state is that it works everywhere apart from the PWA (in normal/mobile view). I also tried on difference phones, same behavior. Also, I have a hunch that the current improvement is temporary and the issue will reappear sooner or later. |
Beta Was this translation helpful? Give feedback.
-
|
It happens in Safari on iOS as well. |
Beta Was this translation helpful? Give feedback.
-
|
I have a build here with a potential fix and logging that will help narrow it down. Keep in mind this build is based on 0.17
|
Beta Was this translation helpful? Give feedback.
-
|
Seeing similar behavior in Chrome browser. It's intermittent but happens enough to be a usability issue. |
Beta Was this translation helpful? Give feedback.
-
|
Try the following to reproduce the bug:
Tested on Arch Linux/Firefox 145.0.1. Maybe make sure you're not using continuous recording. I have many empty gaps in the timeline (motion detected recording). |
Beta Was this translation helpful? Give feedback.
-
|
Looks like opening from explore showed a lightly different issue, will be fixed in the next build |
Beta Was this translation helpful? Give feedback.
-
|
@zolakt @BlueMax Let us know if the latest dev build corrects your issue. |
Beta Was this translation helpful? Give feedback.
-
|
Sure. Sorry haven't tried it yet. I'll try to find some time over the weekend to test |
Beta Was this translation helpful? Give feedback.
-
|
Seems fixed here. Maybe needs more time to make sure (I've started from scratch with new db) but for now it's working properly. Thx! |
Beta Was this translation helpful? Give feedback.
-
|
Last build seems to have fixed the issue for me as well |
Beta Was this translation helpful? Give feedback.
-
|
Closing this as resolved, feel free to reopen if there are still reproducible issues. |
Beta Was this translation helpful? Give feedback.
@zolakt @BlueMax Let us know if the latest dev build corrects your issue.