bypassrg / att

Using Asuswrt-Merlin to bypass AT&T's residential gateway
MIT License
153 stars 36 forks source link

Flow acceleration prevents auth #3

Closed tmaiaroto closed 5 years ago

tmaiaroto commented 5 years ago

The readme talks about the hardware acceleration...but I realized that with CTF + FA ("level 2" I guess) it doesn't authenticate.

If I use CTF only (by enabling something like traffic monitoring), it will authenticate just fine.

Does everyone have CTF + FA?

I was wondering if lack of FA was attributing to a slower speed. Though there's too many factors at play - time of day, contention ratio, etc. I can get 600-700mbs down and 900+ up. Though before, with AT&Ts router, I'd get the full gigabit down. Maybe I just haven't caught it at the right time of day...or maybe FA matters?

tmaiaroto commented 5 years ago

Well, despite not having FA, I can get the full gigabit speed (or pretty darn close at 905d/913u). I just tried from a faster machine. Surprisingly a Surface Pro isn't getting gigabit even with the gigabit Ethernet adapter. I would say I feel cheated for the cost of the USB adapter, but it certainly is faster than wifi and many other USB Ethernet adapters aren't gigabit, they're 100mbps. So I guess the USB Ethernet still has value.

Good to know I don't "need" FA...but regardless, I'm still curious if it's something that should be ok to use and if it leads to more consistent performance benefits.

I have to turn something on to disable FA. Right now I chose some feature under aiprotection. Traffic monitoring would also disable FA and I noticed I could turn off aiprotection (contrary to the readme). ... So is this what's going on? Is it not so much aiprotect doing something as it is the disabling of FA?

tmaiaroto commented 5 years ago

Gonna just close this since it seems to not be an issue for performance? ...but would be good if anyone has any insight here. It's now lower on my list (battling VLANs and DHCP and such), but if I ever figure it out I will comment back here for posterity.