No rtsp stream after h264_mp4toannexb error

[firefoot007]

Hi,

After blingbridge with mediamtx keeps running for 1-2 days, the rtsp stream stops and the following error shows up in the docker log files.

[concat @ 0x7fbc917a7600] h264_mp4toannexb filter failed to receive output packet
[NULL @ 0x7fbc918be980] Invalid NAL unit size (-2023640375 > 20454).
[NULL @ 0x7fbc918be980] missing picture in access unit with size 20458
[concat @ 0x7fbc917a7600] h264_mp4toannexb filter failed to receive output packet
[NULL @ 0x7fbc918be980] Invalid NAL unit size (-2023640375 > 20454).
[NULL @ 0x7fbc918be980] missing picture in access unit with size 20458
[concat @ 0x7fbc917a7600] h264_mp4toannexb filter failed to receive output packet
[NULL @ 0x7fbc918be980] Invalid NAL unit size (-2023640375 > 20454).

It keeps on repeating until I restart the container, and then it starts outputting the rtsp stream. I am using local storage with Sync Module 2.

Secondly, as I mentioned before, the RTSP stream when viewed by both Frigate and VLC is very low fps, like 1/6 of the actual scene fps. Maybe I need to change some ffmpeg stream args? Can I change them in the config file?

And, thank you for sharing this project with us, it makes my doorbell useful.

[roger-]

Weird, I’ve actually had mine running for months and haven’t seen this error. I’m using cloud storage though so maybe there’s another issue there.

I could definitely use some help optimizing the parameters. I’d recommend running it locally (manually running mediamtx) and tweaking the parameters in ffmpeg.py and stream_server.py (the latter I think shouldn’t need changing).

[firefoot007]

Okay, I will monitor again. In the meantime, I have set the mediamtx stream as a restream in Frigate go2rtc to see if it changes it.

An internet search told me that it could be due to a corrupt frame in the video downloaded from the Blink server. I just reset the USB stick connected to the sync module 2, so it does not cause any error.