Packet loss with Billion 8800NL when PC is on & connected?

Hi all,

Recently purchased an Billion 8800nl r1 on ebay and put in bridge mode and updated it to the stable DH65 firmware. Thing is I'm getting quite a bit of packet loss whenever my PC is on & connected and NOT actively using the internet.

Configuration:

Billion 8800NL R1 bridge ---> Asus RT-AC68U routing/wlan

2 ethernet cables run from Billion modem to Asus router:
- one from Billion LAN1 to Asus WAN port
- one from Billion LAN2 to Asus LAN2 port

PC is directly connected to LAN 1 port on Asus. No other ethernet ports in use.

BQM illustrating a specific example of packet loss:
https://www.thinkbroadband.com/broadband/monitoring/...

Steps taken: Tried DH65 firmware and newest firmware also. Factory reset performed after both upgrades. Eliminated Asus router as cause by using HG612 in lieu of Billion modem - no packet loss at all detected by BQM. Did interface grouping to separate LAN ports etc. Disabled firmware on Billion modem and Asus router. Enabled WAN ping on both.

Absolutely stumped as to what could be causing the issue. For information the Billion is a great modem with higher actual/attainable sync than the HG612 and I am on the verge of selling it on because of this issue.

Would be great if anyone could help me troubleshoot this.

I have to R2 in the black casing, without any packet loss, but running in combo mode. See live BQM links in my sig. AIUI the works are the same as yours, just the new casing.

Any chance you could give us the live link for yours please to see if anyone spots anything from that?

Edited by RobertoS (Fri 19-May-17 20:09:26)

I wonder if either the PC or maybe the monitor are injecting RF into the circuit ?

Do you have any way to physically disconnect (and power off + unplug from mains) the monitor, test and then repeat with the PC .. ... just a thought.

Live link to BQM: was running HG612 before disconnecting and connecting the 8800NL up just now (20:45):

https://www.thinkbroadband.com/broadband/monitoring/...

MyDSLWebStats: fishpan

Wouldn't this also affect the HG612 also? I get no packet loss at all with this modem while keeping every other variable the same.

-Edit: The monitor is a Raspberry Pi (headless aka no mouse/kb) which is connected to the Asus Router's LAN 3 port. Power is provided via USB cable from Asus router's USB 3.0 port to RPi's microUSB port. (I thought this could inject some RF perhaps - but HG612 shows no sign of packet loss at all whereas 8800NL shows huge amounts of packet loss.)

Edited by fishpan (Fri 19-May-17 21:08:25)

Ah yes, good point.....

unless somehow the onboard modem is closer to the source ....

do you achieve a better sync rate when using the Asus ? Maybe the HG is plodding and therefore less sensitive ? (More guesses, I'm afraid)

Billion gets around 1-2 mbps more sync rate (72 vs 70) than the HG612. No discernable differences in error rates FEC/CRC/ES/SES/HEC etc between the two modems. Exact same DSL cable (Tandy twisted pair 0.5m) used between NTE5C/Mk4 and the modems so both same distance/position from socket.

I have a vague idea that the PC is somehow overloading the Billion modem by bombarding with packets at certain times such that the Billion modem just stops responding to certain ping packets from TBB. However, I've observed that the packet loss never occurs when I am actively doing internet stuff on the PC, only when I leave the PC idling and not downloading anything.

That's a weird one for sure. Keep us posted.

As Zarjaz said, that's very strange. Could this be a faulty Billion BiPAC 8800NL?

I found that the Billion BiPAC 8800NL R2 didn't perform that well on my line. It connected at a much lower downstream rate than any other router I've tested except the BT "not so Smart" Hub and the downstream SNR margin would go up and down like a yo-yo constantly.

I went for the R1 after reading about how the R2 uses a more "budget" chipset. Minor differences but R1 had a more "mature" chipset that was capable of higher sync on longer lines according to kitz.

Not quite sure what to make of that. the live one looks fine apart from two spikes, one just after midnight morning of today 19th and one later in the day. Then the big one soon after the resync. Following that a couple of weirdos neither of which coincides with the latency spike.

Will check now and then on that live link .

Edit: How odd! I have the same packet loss just after midnight in exactly then same place! On AAISP.

Edited by RobertoS (Fri 19-May-17 22:54:56)

Is your AAISP connection BTW or TTB?

BTW with WLR3 line from Pulse8.

At least, that's what I've always thought.

Edited by RobertoS (Fri 19-May-17 23:39:13)

Agreed!

Live graph now shows packet loss when PC is powered on and nothing is using the internet. Just left the PC on while playing an offline game and this happens:

https://www.thinkbroadband.com/broadband/monitoring/...

I've additionally tried setting the LAN1 port on the 8800NL from 1Gbps to 100Mbps to try and match the 2 lan cables running between the billion modem and the asus router. Hasn't done squat.

Going to shift the LAN cable used for monitoring DSLstats from LAN2 on the billion to LAN3 to try and eliminate the port itself as faulty.

I can't see what is wrong with your link but it gives a 404. Your live one earlier is working fine.

Strange all that packet loss if nothing at all is running. Have you checked what is connecting to your wifi? Straw-clutching but worth checking.

You haven't got a smart TV have you, with someone running catchup/OnDemand?

Think I have found and fixed the problem AT LAST!

https://www.thinkbroadband.com/broadband/monitoring/...

Setup that was causing packet loss was using 2 ethernet cables from modem to router; one for WAN bridge and one purely to put the modem on the same LAN as the other devices connected to router (for stats collection purposes).

I tried removing the ethernet cable used just for stats (LAN 2 on ASUS to LAN 2 on Billion). I could not access the Billion anymore as there was no LAN IP issued for it by the Asus Router. However, I noticed that the packet loss immediately ceased even when my PC was connected! This unplugging event can be referenced at 12pm on the graph in URL above.

I still wanted to retain the stats collection via DSLstats so that my Raspberry Pi could upload them to MWDS. To do this I then followed Chrysalis' post:http://forum.kitz.co.uk/index.php?topic=14621.0

Now I have the Billion modem providing internet AND stats to the Asus router's WAN port via ONE ethernet cable. To access the Billion, I use 192.168.2.254 in my browser (default 2nd subnet LAN IP). To access the Asus, I use 192.168.1.1 which is it's default primary subnet LAN IP. Packet loss has dropped to 0% going by the TBB graph.

I'm not that experienced in the networking quirks of routers but would hazard a guess and say that there was some kind of routing loop that was caused when I used two cables for concurrent internet and DSL stats collection that manifested when my PC also connected via ethernet.

An added note is that I had this same packet loss issue on a Zyxel VMG8924, Zyxel VMG3925 and finally on this Billion 8800NL R1. However, the HG612 was seemingly immune perhaps due to its default setup being configured to avoid this routing loopback?

Hope this helps anyone who might stumble into this issue in future.

-edit: roughly 2PM on the TBB BQM graph is when I used chrysalis' method and implemented the 1 cable config

Edited by fishpan (Sun 21-May-17 19:36:23)

Looks good . Well sorted.

Yes I can see the 2pm drop and clear from then on. Fingers crossed .

I'll also bookmark this in case I decide to use my ASUS router in the same way. As you say, it wasn't a problem with the HG612.

Edited by RobertoS (Sun 21-May-17 19:55:09)

As you correctly guessed you had created a loop between the LAN ports by connecting two ethernet cables between the Billion and the Asus so the switch on the Billion was being "flooded" and overloading the processor.

But the WAN connection is surely dedicated in some way? The WAN link is not on the LAN, so how can there be a loop? Even though the IP Address of the "modem" is on the LAN subnet.

That's how the stats reading on the HG612 is done, by many of us.

I suppose it could be possible to connect the WAN output of a bridged modem/router such as the HG612 to any old ethernet port on the active router and get things working, in the same way as we can daisy-routers. I can see how that coukd set up the loop. I did even wonder if that was what the OP had done.

The Billion and ASUS have nominated WAN ports which can be set in their GUIs to LAN use, presumably avoiding the loop when in WAN mode.

The dedicated WAN port on the Billion 8800NL can't be set when it's in bridge-mode so it's operating as a switch and anything on that comes from the router WAN port appears on the other 3 ports on the 8800NL.

My setup with an 8800NL is: 8800NL --> Sky Hub --> Samknows White box --> 16 port gigabit smart switch. then I have a separate cable from the switch going to another LAN port on the 8800NL for the stats monitoring.

An ethernet port on the Billion is feeding the WAN port on the ASUS, with another cable for the Lan use. The separation from the LAN is surely occurring at the ASUS? The setup the OP described at the start is identical to the HG612 + ASUS configuration which I used for a long time.

Why should one work and not the other?

(I'm not saying you are wrong, I genuinely don't see the difference).

Aren't the two LAN ports on the HG612 separately configurable because I'm sure you can set the LAN2 IP address once it's unlocked, so it would work differently to something like the 8800NL on which the ports can't have separate IP addresses.

I'll have to dig out my old router and connect it up with two cables and see what happens because the Q hub only has 2 LAN ports.

On the HG612you simply set or leave the default LAN address as is. I forget what the out-of-the-or-box one is, but let's say 192.168.0.1, and that the router address is 192.168.0.254. Both could equally be set to 192.168.1.nnn.

That's all. Connect LAN1 which is the WAN link for your router to your router WAN port, and router LAN port back to LAN2 , job done.

Access to the HG612 GUI and telnet interface by ethernet or wireless through the router, on 192.168.0.1.

As the WAN connection of any router ignores the default IP address of the HG612 which remains in bridge mode, no problem. The Lan2 port responds as it is on a fixed IP address on the router's subnet, with DHCP not active on the HG612.

Edited by RobertoS (Mon 22-May-17 16:36:50)

Overloading the Billion modem's CPU makes sense as its GUI would often get very unstable even on the DH65 firmware which is touted by many as having the most stable and fast firmware for the v1 8800nl.

I am glad someone found my notes useful, good you have a working solution now.

I find 2.32e better because it seems to be more stable on long lines.

Thanks for the detailed notes - am a complete newb when it comes to fiddling around in telnet/ssh but your guide made it really easy to follow. Helped that i had the same Asus + Billion setup too.

Have you found a way to make the settings stick on rebooting the router? At the mo I need to input the commands again on reboot otherwise the Asus can't see the Billion and thus MDWS doesn't work.

Thanks

install merlin's build of asuswrt, and then you can add the command to a script to run when the wan comes online.

https://asuswrt.lostrealm.ca/

info on scripts here

https://github.com/RMerl/asuswrt-merlin/wiki/User-sc...