[Support]: API bugs

[skrashevich]

Describe the problem you are having

I received an MQTT event with the following content:

{
    "before": {
        "id": "1683526465.197185-e91hv2",
        "camera": "kitchen",
        "frame_time": 1683527818.17968,
        "snapshot": {
            "frame_time": 1683526468.342398,
            "box": [
                456,
                806,
                711,
                1078
            ],
            "area": 69360,
            "region": [
                381,
                676,
                785,
                1080
            ],
            "score": 0.87109375
        },
        "label": "bottle",
        "sub_label": null,
        "top_score": 0.88720703125,
        "false_positive": false,
        "start_time": 1683526465.197185,
        "end_time": null,
        "score": 0.88720703125,
        "box": [
            458,
            806,
            710,
            1078
        ],
        "area": 68544,
        "ratio": 0.9264705882352942,
        "region": [
            396,
            704,
            772,
            1080
        ],
        "stationary": true,
        "motionless_count": 13339,
        "position_changes": 1,
        "current_zones": [],
        "entered_zones": [],
        "has_clip": true,
        "has_snapshot": true
    },
    "after": {
        "id": "1683526465.197185-e91hv2",
        "camera": "kitchen",
        "frame_time": 1683527878.332771,
        "snapshot": {
            "frame_time": 1683526468.342398,
            "box": [
                456,
                806,
                711,
                1078
            ],
            "area": 69360,
            "region": [
                381,
                676,
                785,
                1080
            ],
            "score": 0.87109375
        },
        "label": "bottle",
        "sub_label": null,
        "top_score": 0.88720703125,
        "false_positive": false,
        "start_time": 1683526465.197185,
        "end_time": null,
        "score": 0.88720703125,
        "box": [
            458,
            806,
            710,
            1078
        ],
        "area": 68544,
        "ratio": 0.9264705882352942,
        "region": [
            396,
            704,
            772,
            1080
        ],
        "stationary": true,
        "motionless_count": 13941,
        "position_changes": 1,
        "current_zones": [],
        "entered_zones": [],
        "has_clip": true,
        "has_snapshot": true
    },
    "type": "update"
}

I see that the has_clip field is true. This means I can get the video of this event.
I take the after.snapshot.frame_time field (1683526468.342398), as well as before.start_time (1683526465.197185) and try to get the video between these timestamps through the API. Just in case, I add 5 seconds on both sides.
I form a request to the API: curl http://localhost:5001/kitchen/recordings?before=1683526460&after=1683526468
I get the response: []
What I'm doing wrong?

Version

latest dev

Frigate config file

record:
  enabled: True
  expire_interval: 60
  retain:
    days: 10
    mode: motion
  events:
    pre_capture: 30
    post_capture: 120
    retain:
      default: 7
      mode: motion
      objects:
        person: 180
        bottle: 1
        car: 7

snapshots:
  enabled: True
  height: 1080

Relevant log output

2023-05-08 09:54:12.133335604  x.x.x.x - - [08/May/2023:09:54:12 +0300] "GET /api/kitchen/recordings?before=1683526460&after=1683526468 HTTP/1.1" 200 3 "-" "curl/7.87.0" "-"

FFprobe output from your camera

unrelevant

Frigate stats

No response

Operating system

Other Linux

Install method

HassOS Addon

Coral version

USB

Network connection

Wired

Camera make and model

camera-independent-issue

Any other information that may be helpful

No response

[NickM-27]

has_clip simply means that the camera currently has recording enabled, it's not an indication that the clip is ready. The segment hasn't finished processing from cache most likely.

#2128 is a related feature request

[skrashevich]

My task is obtain a video clip of the minimum possible length (but not less then -5:+5) within the range of -5:+5 seconds concerning the moment of event detection. Do you have any suggestions regarding a workaround?

[NickM-27]

you'll need to give it some delay between when the message comes through and your request to get the recording

[skrashevich]

you'll need to give it some delay between when the message comes through and your request to get the recording

I had hoped that the presence of the attribute "has_clip" means the existence of the record, rather than its potential availability 🙁

[NickM-27]

I had hoped that the presence of the attribute "has_clip" means the existence of the record, rather than its potential availability 🙁

Right, but it doesn't, which is the reason for the pinned feature request I linked above

[skrashevich]

Right, but it doesn't, which is the reason for the pinned feature request I linked above

Do you have any plans to make significant changes to the recording manager subsystem (#6248) in the near future? I have some ideas on how to resolve these issues and implement flexible recording storage settings, but I would prefer not to do parallel useless work.

[NickM-27]

@blakeblackshear would be better to answer that question

[blakeblackshear]

There aren't any additional plans to overhaul the recordings in the next release beyond what has already been merged into dev. I would definitely suggest outlining your ideas ahead of time in case I think they are problematic long term.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.