Ubuntu 16.04 & MLPPP
-
I know this will likely go over a lot of people's heads, but I've got an interesting thing happening, and I'm at a total loss to explain it.
To start, I'll point out that my server configuration is a little different than the average person's, for I have the following setup:
First off, I'm running Ubuntu 16.04.1 x64. The server currently has three network cards. Their configuration is as follows:
eth0 - 1000mbit to a 5-port switch - Modem #1 & main WiFi WAN connect at 100mbit each
eth1 - 100mbit directly to Modem #2
eth2 - 1000mbit to an 8-port switch - Modem #3 connects to the switch at 100mbit, and the trailer's WAN connects to the switch at 1000mbit
eth3 - 100mbit direct to Modem #4 (Coming Soon)Lines 1, 2, & 3 are currently serviced with a pre-Y2K non-FTTN legacy DSL service, running on a 4032/800 fastpath profile (although they're getting upped on Tuesday to 5056/800 interleave). When joined together, I'm getting just over 10mbit down, about 1.85mbit up. Modem 1, 2, & 3 in the equation (which are the only ones with service currently) are identical TP-Link TD-8816 modems, and the 4th modem that will soon be added into the mix is a Thompson SpeedTouch 516.
Now here's where things get interesting… The server bridges all 3 modems together, and when connected through all 3 modems, I can use Traceroute directly from the server with no issue. I get the following:
noel@mail:~$ traceroute teksavvy.com traceroute to teksavvy.com (173.246.155.216), 30 hops max, 60 byte packets 1 206.248.154.121 (206.248.154.121) 22.423 ms 22.383 ms 22.612 ms 2 ae0-2150-bdr01-tor.teksavvy.com (69.196.136.172) 22.597 ms 22.833 ms 22.819 ms 3 ae8-0-bdr01-tor2.teksavvy.com (206.248.155.9) 22.805 ms ae1-0-agg01-tor2.teksavvy.com (206.248.155.14) 23.038 ms 23.023 ms 4 ae0-2170-agg01-tor2.teksavvy.com (192.171.59.98) 23.510 ms cdn.teksavvy.com (173.246.155.216) 23.485 ms ae0-2170-agg01-tor2.teksavvy.com (192.171.59.98) 23.723 ms noel@mail:~$As well, in ICMP mode, I get the following
noel@mail:~$ traceroute -I teksavvy.com traceroute to teksavvy.com (173.246.155.216), 30 hops max, 60 byte packets 1 206.248.154.121 (206.248.154.121) 9.762 ms 10.225 ms 11.364 ms 2 ae0-2150-bdr01-tor.teksavvy.com (69.196.136.172) 13.096 ms 13.911 ms 15.086 ms 3 ae1-0-agg01-tor2.teksavvy.com (206.248.155.14) 16.359 ms 17.200 ms 20.284 ms 4 cdn.teksavvy.com (173.246.155.216) 19.434 ms 20.696 ms 21.418 ms noel@mail:~$Now here's where things get a little interesting. When I am on say my laptop connecting through the network then out to the same DSL lines through the server, I get these same results (with a couple of extra hops). As soon as the server has established a PPPoE link through modem 2 and/or modem 3 however, things get a little strange (as in this example):
noel@New-Laptop:~$ traceroute teksavvy.com traceroute to teksavvy.com (173.246.155.216), 64 hops max 1 192.168.100.1 (router.trailer.ohsweken) 0.365ms 0.608ms 0.469ms 2 192.168.0.1 (server.ohsweken) 0.656ms 0.565ms 0.482ms 3 * * * 4 * * * 5 * * * ^C noel@New-Laptop:~$What gets more interesting, is that when I add ICMP mode to the mix, it works just fine, as in this example:
noel@New-Laptop:~$ traceroute -I teksavvy.com traceroute to teksavvy.com (173.246.155.216), 64 hops max 1 192.168.100.1 (router.trailer.ohsweken) 4.529ms 0.391ms 0.316ms 2 192.168.0.1 (server.ohsweken) 0.633ms 0.635ms 0.620ms 3 206.248.154.121 (206.248.154.121) 11.241ms 10.595ms 9.976ms 4 69.196.136.172 (ae0-2150-bdr01-tor.teksavvy.com) 10.934ms 9.878ms 10.258ms 5 206.248.155.14 (ae1-0-agg01-tor2.teksavvy.com) 16.042ms 11.698ms 9.858ms 6 173.246.155.216 (cdn.teksavvy.com) 10.114ms 11.691ms 10.520ms noel@New-Laptop:~$Does anyone have a best guess (or even better, actually know) as to why this might be? I'm very confused and it's driving me absolutely bat shit crazy!
Please help… If you can.
-
Hmmmm, technical smart guy :cool2:, I wanna jump in bed with you >:D!
-
Where there's a will to get 1080p HD video streams, there's a way to make it possible, even on archaic, crappy, oldschool DSL lines lol
-
So here's another interesting question along the same lines as the other one….
I've added a third line to the MLPPP setup, and I'm having a hell of a time getting Ubuntu to play nice with bonding all 3 lines... Anyone have any advice or should I just bite the bullet and reformat the server to make it work?
-
Cool pic MrMazda!
Does the laptop connect to the server or to linkboxes? Are efficiency and exposure reasons to lose few extra steps or?
–
Btw, I have a bunch of adsl2/vdsl2 modems that I could donate away.
-
I ended up solving the problem by getting a little creative. The three DSL modems now wire directly into a Mikrotik RB-750 router, which establishes the PPPoE session through the 3 modems and bonds them together, then shares the resulting connection with my server. I had to get a little more creative because I host a static /28 subnet on my server, which used to handle the connection, since the server itself would choke when I try 3 lines at the same time.
So… The router establishes the PPPoE session and receives the 69.xxx.xxx.6 IP address. From there, the router hosts the /30 subnet using 69.xxx.xxx.177, with the server on 69.xxx.xxx.178. I had to add in a routing rule because my ISP will only push the next hop for all my routing to the IP address of the PPPoE link. So, a few tweaks later to add a routing rule to tell it that if the destination is 173.xxx.xxx.80/28, forward the packet to the server at 69.xxx.xxx.178 as the next hop. Once I got that all setup, I was back in business.
As for my laptop, it's a little more creative. I now only use two network cards in my server. The first network links into the Mikrotik router in which to get its internet connection, then shares it on the second network card. The wiring then leads from the second network card to a 16 port 1000mbit switch, which all the devices in the house hook into. Also, this main switch is where I have another line link it that leads out through the outside wall, down into the ground, where it then runs about 200' or so underground to the trailer. It then magically reappears above ground behind the trailer and leads inside to the WAN port on a Linksys WRT610N router mounted to the inside of the one wall. This router is separate from the D-Link DIR-605L router that sits on top of the entertainment stand in the house to run the main WiFi.
It's a little complicated by design, but at the end of the day, I was more concerned with making sure THAT everything works, including the ability to stream seamlessly in 1080p HD, rather than focusing too much on HOW everything works.
See…. Out here, we only have pre-Y2K legacy DSL on copper. We don't even have FTTN (Fibre To The Node) out here because the Oligopoly sucks. So... Because of this and because of the fact that I am a little more than 5kms away from the SLAM, the best that I can get and remain stable with is a 4032/800 profile, which only gives me about 3.2 to 3.5 mbit down, and a mere 0.6 mbit up. Once the Mikrotik router bridges the three DSL modems together using MLPPP, I magically end up with a connection that runs on a single line that can handle between 10.2 and 11.1mbit down (depending on network load levels in the neighbourhood), and consistently gives me around 1.75mbit up.
As for the laptop and all the handheld devices, they connect VIA WiFi, except there's a bit of a catch... The connection is routed through the Linux box as my firewall, which then if permitted, allows the connection to pass through to the Mikrotik router, where it is then split into three different segments before leaving on the 3 DSL modems, only to be reassembled on the other end by my ISP. Even the DHCP is run through the server, so quite literally, if the server goes down (which it did recently), EVERYTHING stops working, including the home automation system which allows us to do things like lock or unlock the front door from our phone, check out the IP cameras, change the thermostat settings on the furnace, etc. hehe... It's a very custom setup indeed.
