This page says that you may experience a blank screen if certain ports are blocked by the firewall... Is it possible that SQM is either blocking ports are causing the UDP traffic to get blocked?
Okay, this indicates that your kernel automatically adds 14 bytes of the overhead, so you should specify 40-14 = 26 Bytes to sqm (sqm has currently no way to measure how much overhead the kernel already adds to the packet size so this correction needs to be done manually). This is true for both ingress and egress.
This is also orthogonal to your Ring issue, but still you might as well fix this while we are at it
Ah, you could try to test that by setting upload/egress bandwidth in the sqm GUI to zero (which disables the shaper in that direction) if that gets you the video back it would indicate a egress bandwidth issue. You could also try the same with the download/ingress bandwidth for completeness.
This seems literally correct, but since the OP seems to get working video streams without sqm there might be a way. Also the Ring numbers 1 or 2 Mbps really look like someone applied generous rounding to get to nice numbers and not so much like the real technical limits.... (but without any real data, I might be deluded and the technical limit truly is 1Mbps).
I get your point, but is seems to me the OP was sort of accepting that that might be the best he could hope for and was more puzzled that with sqm he did not even get that... I could be wrong, since I can not read anybodies mind ;), but this is getting a bit off-topic, so I will try to refrain from further tangents (which I am just too prone to follow).
Yeah, I have accepted the poor performance... but it works well enough to know if I actually know the person at my door or not, and I could make out what they are saying. But poor performance and not working at all are worlds apart.
I have given Ring feedback that it would be a nice feature if the doorbell was able to livestream a low quality stream, while locally buffering the higher quality video. Once the live interaction is complete, upload the high quality stream. They probably wont do that, since apparently my nieghborhood is the only in the USA with such poor internet access.
When I set out to implement QoS in the router, I was going to do it the old school way. Prioritize the Ring over all other traffic, but I would still suffer from BufferBloat for our more interactive uses which makes me sad.
Is there a way that you can think of where I can apply SQM to everything except the Ring device?
Technically on egress, and that is what we talking about here, the bandwidth sacrifice is really minimal, with proper link layer adjustments one should be able to almost reach 100% of line rate, but still a managed buffer is going to behave differently from a non-managed one... @av_boy the exclude anything from traffic shaping idea unfortunately does not work, once you allow un-managed flows the bandwidth availabe to the shaper gets time vaiant and sqm is not equipped to deal with that.
But there might still be a few tricks up our sleeves to trade some more latency for more tolerable Ring video.
It might make sense to try to increase the interval and target fields in the GUI....
Not really, what it would do however is choke all other traffic while ring is streaming, allowing your neighbourgs to accidentally take you off the internet (most downloads require some feedback via uploaded packets, if those can not go out the download will drizzle out and stall).
It wouldn't surprise me if what happens is the lack of bufferbloat allows the ring device to determine it doesn't have the bandwidth it needs, and it turns off the video, whereas with bufferbloat it just saturates the buffers with video and never sees a drop, so it doesn't know that it's operating on a link that's too slow.
could just be the default fair queueing which will give other concurrent flows as much chance/bandwidth as the Ring, in that case he might try the "besteffort" keyword, but in that mode sqm is not going to be too pleasant to use
I wanted to follow up and thank you for your help trying to fix my issue. I agree that its probably likely that bufferbloat is the reason why the ring is working without SQM. Looking at the minimum specs for the ring, its actually 2mbps for the version I have, not 1mbps. So, I agree that my internet capability is just too crappy to support that device.
I used the information you guys provided as well to tune SQM:
assuming standard PPPoE on the wire size of a packet: 53*(ceil((1500 + 6 + 6 + 2 + 18) / 48)) = 1696
It i s a bit sobering that sqm and Ring seem to be incompatible, but it also seems that you now have bufferbloat in your uplink well under control, so half-full or half-empty, you decide
