Issues with API - specifically the URL

[Normanras]

Up until a few days ago, Frigate had been working really, really well. The particular way I was using it - and now issue - is via Telegram commands and notifications. In particular I a few commands that would send me the last or best picture. The URL being taken straight from documentation:
http://ip-address:5000/api/location/object/best.jpg or http://ip-address:5000/api/driveway/latest.jpg.

Not sure what happened, but the commands still send me a picture, but most of them are just black. If I go into the media library in Home Assistant, the pictures exist and I see the one that should have been sent. In the Frigate UI, the small icon for each object (that previews the image) is black, but when I open up the list of events and pictures, I see all the pictures there. The same pictures as in the media library.

I don't think (yet) that this is a bug, but perhaps something I've done and I wanted to see if anyone else has experienced this. The only changes that have happened in the last few days that I think could have caused this:

• HA updates to 8.6, 8.7, and then tonight 8.8
• I renamed the entity ids of a few automation that use either the commands or send me pictures after a MQTT message (which also doesn't work now). I shortened the entities to automation.a_really_long_id to automation.location_object

I don't believe I touched my YAML file and just looked over it before posting... it seems good. If anyone has any thoughts, please let me know!

[ozett]

i have "best.jpg", but not "last.jpg"

[Normanras]

Apologies - it should say "latest". I will update my post. Also, the latest URL would be without the object. It will just pull a current snap shot from that camera, even if nothing has triggered it.

[blakeblackshear]

latest.jpg should only be black briefly on startup before the first frame has made it to frigate. best.jpg is expected to be black until the next object is detected after a restart.

Is this happening consistently? I have seen this happen occasionally, but it has been rare. Anything in the logs? I don't believe this has anything to do with the HA updates.

Of note, I think most people have moved over to using the frigate/events mqtt topic and fetching snapshots/thumbnails using the specific event id. The best.jpg endpoint is really leftover from a time before objects were tracked.

[Normanras]

Wow, yes lots of errors. I can't believe I didn't check these sooner. They seem to repeat themselves in this block:

frigate.video INFO : porch: ffmpeg sent a broken frame. memoryview assignment: lvalue and rvalue have different structures frigate.video INFO : porch: ffmpeg process is not running. exiting capture thread... ffmpeg.porch.detect ERROR : [flv @ 0x556ee254a0] Timestamps are unset in a packet for stream 0. This is deprecated and will stop working in the future. Fix your code to set the timestamps properly ffmpeg.porch.detect ERROR : av_interleaved_write_frame(): Connection reset by peer ffmpeg.porch.detect ERROR : Finishing stream 1:0 without any data written to it. ffmpeg.porch.detect ERROR : [flv @ 0x556ee254a0] Failed to update header with correct duration. ffmpeg.porch.detect ERROR : [flv @ 0x556ee254a0] Failed to update header with correct filesize. ffmpeg.porch.detect ERROR : Error writing trailer of rtmp://127.0.0.1/live/porch: Connection reset by peer

I don't think any of this info is sensitive, but you would know best @blakeblackshear. I'll edit out if you see anything that shouldn't be public.

I don't have any ffmpeg rules in my yaml except the basic of declaring ffmpeg - no extra parameters. I think I'll have to pare down my frigate file and see if anything changes. Start with the basics.

[ozett]

i had mostly to check and adjust height and width of stream and in the config for similar error-msg..
after that i checked the ffmpeg params... good luck

[blakeblackshear]

I don't know what type of camera you have, but these errors are expected to happen occasionally unless you have very stable PoE cameras. Normally, frigate will restart ffmpeg and recover.

[Normanras]

Thanks all - yeah height and width and all other parameters have been functioning flawless (you're truly amazing, blake!) other than this random glitch. I would say I have stable PoE cameras, but I can't be 100% sure. I've already rebuilt the add-on, but let me change the yaml and restart the add-on. Hopefully, that will fix it.