| 1 |
>> > > If that device behaves badly in router mode by blocking just all |
| 2 |
>> > > icmp traffic instead of only icmp-echo-req, this is a good idea. |
| 3 |
>> > > You may want to bug AT&T about this problem then. It should really |
| 4 |
>> > > not block related icmp traffic. |
| 5 |
>> > |
| 6 |
>> > |
| 7 |
>> > Hi Kai, yesterday I switched my Gentoo router over to handling PPPoE |
| 8 |
>> > and pings seem to be working properly now. The AT&T device is now |
| 9 |
>> > functioning as a modem only and passing everything through. Today |
| 10 |
>> > I'll find out if it helps with TCP Queuing and (supposedly) related |
| 11 |
>> > http response slowdowns. |
| 12 |
>> |
| 13 |
>> You may want to set the default congestion control to fq-codel (it's |
| 14 |
>> in the kernel) if you're using DSL links. This may help your problem a |
| 15 |
>> little bit. It is most effective if you deploy traffic shaping at the |
| 16 |
>> same time. There was once something like wondershaper. Trick is to get |
| 17 |
>> the TCP queuing back inside your router (that is where you deployed |
| 18 |
>> pppoe) as otherwise packets will queue up in the modem (dsl modems use |
| 19 |
>> huge queues by default). This works by lowering the uplink bandwith to |
| 20 |
>> 80-90% of measured maximum upload (the excess bandwidth is for short |
| 21 |
>> bursts of traffic). Traffic shaping now re-orders the packets. It |
| 22 |
>> should send ACK and small packets first. This should solve your |
| 23 |
>> queuing problem. |
| 24 |
>> |
| 25 |
>> Between each step check dslreports.com for bufferbloat. I'm guessing |
| 26 |
>> it is currently way above 1000 ms while it should stay below 20-50 ms |
| 27 |
>> for dsl. |
| 28 |
>> |
| 29 |
>> The fq-codel congestion control fights against buffer bloat. But it |
| 30 |
>> can only effectively work if you're doing traffic shaping at least on |
| 31 |
>> your uplink (downlink may or may not be worth the effort depending on |
| 32 |
>> your use-case). |
| 33 |
>> |
| 34 |
>> Additionally, you can lower the priority of icmp-echo-reply this way |
| 35 |
>> so during icmp flooding your uplink will still work. |
| 36 |
>> |
| 37 |
>> This link may help you: |
| 38 |
>> https://www.bufferbloat.net/projects/codel/wiki/Cake/ |
| 39 |
> |
| 40 |
> And this: |
| 41 |
> https://github.com/tohojo/sqm-scripts |
| 42 |
|
| 43 |
|
| 44 |
I haven't mentioned it yet, but several times I've seen the website |
| 45 |
perform fine all day until I browse to it myself and then all of a |
| 46 |
sudden it's super slow for me and my third-party monitor. WTF??? |
| 47 |
|
| 48 |
- Grant |
Archives