Ghost QOS or SQM, stuck at SLOW

Ah, let me concur with @tapper, it makes a lot of sense to tackle problems like this segment by segment. Would you be willing to repeat these experients with your laptop connected via an ethernet cable? This would avoid having to deal with your uplink and your wifi at the same time?
Also, depending on how much configuration you did (and whether you actually made a configuration back-up before the sqm/qos experiments) you could use the "firstboot" command from an ssh login on the router to undo all changes from the pristine state, you then will either have to manually configure your router again or use the backup/restore page in the GUI to restore a backup you made.

BTW. the bufferbloat lots of all three tests you posted are impressively bad/ badly impressive, > 7 seconds of bufferbloat is atrocious.

Silly idea, rerun the speedtest while also using "mtr 8.8.8.8" on your laptop concurrently and then see wheher yu can find a hop from which on RTTs are terrible (mtr/traceroute rasults are typically a bit harder to read than one would like, but having RTT massively increase from a given hop on is still a pretty clear piece of information of something fishy happening at that hop). While not beng an expert, I would happily comment upon the mtr results if you post them here (and since when was "not-being-an-expert" sufficient rationale for not having/voicing an opinion :wink: )

(Sorry for the radio silence. As a newly registered user, I hit an anti-spam quota for postings, so I had to wait a while to reply.)

After seeing those preliminary hard-wired results, I continued with that connection. My bufferbloat minimized at about 450 mbps download, which is pretty good when you consider that I am supposed to be getting only 50 mbps. (I suspect that I am provisioned for 50 or a little over during congested times, and I'm allowed to burst above that when there is idle bandwidth.)

Thanks to all for the many comments and suggestions. I think we can chalk up the initial problem to coincidental timing of odd behavior over the wifi testing.

Or your ISP's configuration is confused (maybe thinking you got a 500/15 plan :wink: ), so unless you can actually make good use out of the transiently available bandwidth you could call them and ask them to either commit to 500/150 or at least configure 50/15 in a reliable and predictable way. Not sure about frontier, but if this is GPON your egress is controlled by both the ISP's OLT and your ONT, while the ingress is controlled by the OLT (or maybe something higher in the ISPs network). So this could indicate something broken upstream and good luck trying to de-bloat the buffers of a broken/degrading piece of equipment. On the other hand if you can shape to say 100/15 and retain A/A+ bufferbloat rating, maybe just accepting the IPS's nice service while it lasts might also be a decent choice.

Yeah, I'm going to let it ride for a while and see how it goes. :slight_smile: