On the main status page. It should be eth0.12, which is what needs to be set as the interface in the SQM config. Also you need to change enabled from 0 to 1.
In the line config queue eth0, eth0 is just a friendly name you can change it to something that makes more sense like 'wan'
Edited 2;
The file being changed when I change any configuration for SQM is /etc/config/sqm. The other file that was created/downloaded when I installed luci-app-sqm is not being chnaged.
@HelloShitty You can name the interfaces whatever you want. I’ve kept it to the default. This is what SQM setup looks like via GUI (just replace the interface name toggle with your WAN interface name - inputted 90% of your speed as a starting point):
By the way, which file should I replace the default DNS servers? I want to try different ones other than ISP defaults. I can see in luCI DNS Forwardings and in Docs I see this option but also DNS Provider.
Which should be the file to change? I tried to add Cloudflare DNS to luCI in DNS Forwardings but it is not the result I expected. At least I tried to check in some site what was the DNS server in use and it still detected my ISP's one!
Still haven't tested. Waiting for wife and daughter to head to bed and then I'll disconnect all devices from wireless and will test. That's how I conducted the first tests.
Well, after setting Download/Upload thresholds in SQM to 90/90 of what my ISP says that I should get (200/100), I get this result:
I'm a little disappointed with speed but that thing of bufferbloat seems to have improved.
Also, I'm should be checking latency. Where do I check that?
Edited;
I guess I messed up previous tests.
Links are no longer available. Crap. Did anyone saved those images?
Edited 2;
I tried to install ddns-scripts to be able to use dynamic DNS services but I got this output from
Expected output as long as you have not configured the ddns service.
Doing config manually by editing etc/config/ddns is possible, but might be easier in LuCI with luci-app-ddns installed.
The speed is based on the top speed vs others on the same technology, in the same country:
top 2% - A+
top 10% = A
top 20% - B
top 50% - C
top 80% - D
otherwise : F
Note that whether the grade is "good" or "bad" is open to interpretation. Higher speeds and therefore higher grades generally cost more. You may get an "A" or "A+" and be paying too much. You may get an "F" and yet have a good value connection.
I would try out fq_codel + simplest and turn off the advanced settings. See how your results compare. With more aggressive CPU settings the r7800 I can get it to shape up to ~500mbps with fq_codel. I can only get ~low 200’s with cake.
The latency is the bufferbloat timings we see in DSL reports? What about ping time we usually see in these kin of tests?
@ACwifidude I'll have to try that latter. I want to conduct all testing under the same conditions. Meaning, no wireless devices consuming bandwidth during the test. So, I'll run that later.
Previous response from ACwifidude is pretty well detailed and complete about speed, so just adding to that for my part.
I'm shaping very low bandwidth and for me, I don't think that it makes alot of changes from the base. But I subjectively find the router itself to be more responsive. Got me from A to steady A+ everywhere most of the time. Except when I post the results .
Since I did bad trying hnyman's firmware, I'm not very afraid of trying anything. I'm very afraid of losing internet access because of my daughter's classes in the morning. Anyway, I think these settings won't make me lose connectivity to the router, I hope. I mean the SQM settings and the setting - 35 for up_threshold and 10 for sampling_down_factor. Hope they don't make me lose wireless connectivity.
I'm going to read about that threshold and sampling down factor.