NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Mooose
Jan 30, 2019Luminary
Massive packet loss over Ethernet backhaul
I believe I have read all the threads on this topic and tried the suggested solutions that are relevant to my particular network. My apologies if I have missed something!
My Orbi RBK50 (one router, one satellite) connected through the router WAN port directly to the fiber-optic to Ethernet converter supplied by my ISP has been reasonably stable lately.
However, I recently implemented Ethernet backhaul and suddenly Infuse Pro on my Apple TV would stop in the middle of playing video stored on my Drobo every 10-20 minutes.
This was my original setup:
After some initial testing I concluded there were massive amounts of packet loss (some tests up to 30%) happening somewhere.
I can detail all the things I tried to narrow the issue down to the Ethernet backhaul if it is interesting, but I finally ended up trying the most basic setup I could imagine:
Both laptops have wifi disabled.
I still see massive packet loss between the two laptops, and also from a device connected to wifi through the router to the laptop wired to the satellite:
64 bytes from 10.0.0.36: icmp_seq=1572 ttl=128 time=8.543 ms
64 bytes from 10.0.0.36: icmp_seq=1573 ttl=128 time=4.581 ms
64 bytes from 10.0.0.36: icmp_seq=1574 ttl=128 time=3.300 ms
Request timeout for icmp_seq 1575
Request timeout for icmp_seq 1576
Request timeout for icmp_seq 1577
Request timeout for icmp_seq 1578
Request timeout for icmp_seq 1579
Request timeout for icmp_seq 1580
Request timeout for icmp_seq 1581
Request timeout for icmp_seq 1582
Request timeout for icmp_seq 1583
Request timeout for icmp_seq 1584
Request timeout for icmp_seq 1585
Request timeout for icmp_seq 1586
Request timeout for icmp_seq 1587
Request timeout for icmp_seq 1588
Request timeout for icmp_seq 1589
Request timeout for icmp_seq 1590
Request timeout for icmp_seq 1591
Request timeout for icmp_seq 1592
Request timeout for icmp_seq 1593
Request timeout for icmp_seq 1594
Request timeout for icmp_seq 1595
Request timeout for icmp_seq 1596
Request timeout for icmp_seq 1597
Request timeout for icmp_seq 1598
Request timeout for icmp_seq 1599
Request timeout for icmp_seq 1600
Request timeout for icmp_seq 1601
64 bytes from 10.0.0.36: icmp_seq=1602 ttl=128 time=3.282 ms
64 bytes from 10.0.0.36: icmp_seq=1603 ttl=128 time=3.505 ms
64 bytes from 10.0.0.36: icmp_seq=1604 ttl=128 time=3.237 ms
I have unchecked "Enable Daisy-Chain Topology", no difference.
I have applied new firmwares as they have been released, currently both are at V2.2.1.21.
If I unplug the Ethernet backhaul cable (which I have replaced with a short high-quality Cat 6 cable for the test) and restart the satellite and wait for the 5G backhaul to be established I can ping both 10.0.0.33 and 10.0.0.36 with little or no packet loss from the wifi, and zero loss between the two wired laptop wired laptops.
--- 10.0.0.36 ping statistics ---
2078 packets transmitted, 2078 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.651/3.443/23.380/1.186 ms
Ping statistics for 10.0.0.33:
Packets: Sent = 2075, Received = 2075, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 1ms, Maximum = 13ms, Average = 2ms
I can't think of any other explanation other than that the Ethernet backhaul is at fault. I've tried multiple devices connected in different ways and multiple cables connected to different ports.
Simply disconnecting the Ethernet backhaul and leaving everything else the same greatly alleviates the problem.
However, I really want to use Ethernet backhaul as it decreases my latency for the Nintendo switch (included in the "Stuff" normally wired to the satellite) by 20-40%.
Any suggestions?
Mooose Good troubleshooting and interesting observations. It's clear that the packet loss is directly related to the wired backhaul.
I'm now running the ongoing beta software (2.3.0.23) and I can see that NG has done quite good improvements on wired device handling. Would you want to test this SW and see if you will see improvements? If yes, please PM to ChristineT and ask to get the SW download links.
39 Replies
- FURRYe38Guru - Experienced User
Do you get any packet loss with the Laptop directly connected to the Orbi router. Disconnect the satellite from the configuration.
Confirming that your using v2.2.1.210 FW?
How was the satellite connected to ethernet? After the satellite was connected wifi?
For wired back haul:
Set up IP address reservations for each satellite and devices on the router as you add them to the router. The satellites need to be set up via wireless first. Then connect 1 satellite at a time to the ethernet LAN cable. Wait 5 minutes and the top led on the satellite should turn on BLUE. Set up an IP address reservation for the 1st satellite. Continue to do the same thing for the 2nd satellite. Then follow up with your devices.Was the FW updated via auto update or manually?
If auto update, you might do this, re-load the FW file on to both units manually. After loading do a full factory ERASE on both units and setup from scratch. Add the satellite wireless first, then connect the ethernet cable.
Mooose wrote:
I believe I have read all the threads on this topic and tried the suggested solutions that are relevant to my particular network. My apologies if I have missed something!
My Orbi RBK50 (one router, one satellite) connected through the router WAN port directly to the fiber-optic to Ethernet converter supplied by my ISP has been reasonably stable lately.
However, I recently implemented Ethernet backhaul and suddenly Infuse Pro on my Apple TV would stop in the middle of playing video stored on my Drobo every 10-20 minutes.
This was my original setup:
After some initial testing I concluded there were massive amounts of packet loss (some tests up to 30%) happening somewhere.I can detail all the things I tried to narrow the issue down to the Ethernet backhaul if it is interesting, but I finally ended up trying the most basic setup I could imagine:
Both laptops have wifi disabled.I still see massive packet loss between the two laptops, and also from a device connected to wifi through the router to the laptop wired to the satellite:
64 bytes from 10.0.0.36: icmp_seq=1572 ttl=128 time=8.543 ms
64 bytes from 10.0.0.36: icmp_seq=1573 ttl=128 time=4.581 ms
64 bytes from 10.0.0.36: icmp_seq=1574 ttl=128 time=3.300 ms
Request timeout for icmp_seq 1575
Request timeout for icmp_seq 1576
Request timeout for icmp_seq 1577
Request timeout for icmp_seq 1578
Request timeout for icmp_seq 1579
Request timeout for icmp_seq 1580
Request timeout for icmp_seq 1581
Request timeout for icmp_seq 1582
Request timeout for icmp_seq 1583
Request timeout for icmp_seq 1584
Request timeout for icmp_seq 1585
Request timeout for icmp_seq 1586
Request timeout for icmp_seq 1587
Request timeout for icmp_seq 1588
Request timeout for icmp_seq 1589
Request timeout for icmp_seq 1590
Request timeout for icmp_seq 1591
Request timeout for icmp_seq 1592
Request timeout for icmp_seq 1593
Request timeout for icmp_seq 1594
Request timeout for icmp_seq 1595
Request timeout for icmp_seq 1596
Request timeout for icmp_seq 1597
Request timeout for icmp_seq 1598
Request timeout for icmp_seq 1599
Request timeout for icmp_seq 1600
Request timeout for icmp_seq 1601
64 bytes from 10.0.0.36: icmp_seq=1602 ttl=128 time=3.282 ms
64 bytes from 10.0.0.36: icmp_seq=1603 ttl=128 time=3.505 ms
64 bytes from 10.0.0.36: icmp_seq=1604 ttl=128 time=3.237 msI have unchecked "Enable Daisy-Chain Topology", no difference.
I have applied new firmwares as they have been released, currently both are at V2.2.1.21.
If I unplug the Ethernet backhaul cable (which I have replaced with a short high-quality Cat 6 cable for the test) and restart the satellite and wait for the 5G backhaul to be established I can ping both 10.0.0.33 and 10.0.0.36 with little or no packet loss from the wifi, and zero loss between the two wired laptop wired laptops.
--- 10.0.0.36 ping statistics ---
2078 packets transmitted, 2078 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.651/3.443/23.380/1.186 msPing statistics for 10.0.0.33:
Packets: Sent = 2075, Received = 2075, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 1ms, Maximum = 13ms, Average = 2ms
I can't think of any other explanation other than that the Ethernet backhaul is at fault. I've tried multiple devices connected in different ways and multiple cables connected to different ports.Simply disconnecting the Ethernet backhaul and leaving everything else the same greatly alleviates the problem.
However, I really want to use Ethernet backhaul as it decreases my latency for the Nintendo switch (included in the "Stuff" normally wired to the satellite) by 20-40%.
Any suggestions?
- CrimpOnGuru - Experienced User
This is probably a silly question. There are four ethernet ports on both the RBR50 and RBS50. Did you plug the CAT6 ethernet cable into the same ports that were previously used for ethernet backhaul? Do you get the same packet drop when using other combinations of ports?
- MoooseLuminary
There is no packet loss for the laptop (10.0.0.33) wired directly to the router when the using 5G backhaul or no satellite, but with the Ethernet backhaul even this laptop loses packets. See ping results below for details.
Of course, FW V2.2.1.210, copy/paste error, sorry!
For the test below, the satellite was connected to Ethernet after the 5G test, i.e. it was up and running perfectly prior to connecting the cable.
All wired devices, including the satellite, have IP address reservations.
The firmware has been updated by clicking the update button in the web interface.
I have factory-reset both units previously in an attempt to solve a different problem, although I have not downloaded the firmware file from the support page and loaded it, nor have I tried factory-resetting this time around. Is this voodoo, or is there some technical explanation why this may work? I am reluctant to do this as I have IP address reservations and port-forwarding set up the way I need it, so I would really like to avoid this route if there is anything less time-consuming I could try first.
Not a silly question at all, switching ports has solved many issues for me in the past. Regrettably not this time though, and yes, I have tried.
So, here's the test I tried this evening. I started with 5G backhaul:
From the laptop (10.0.0.19) connected to the router through wifi with 5G backhaul:--- 10.0.0.36 ping statistics ---
10000 packets transmitted, 9999 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 2.493/7.358/251.367/9.624 ms--- 10.0.0.33 ping statistics ---
10000 packets transmitted, 10000 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.029/4.328/317.584/8.420 ms--- 10.0.0.2 ping statistics ---
10000 packets transmitted, 10000 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 2.134/7.446/793.257/19.222 ms
From the laptop (10.0.0.33) Ethernet-wired to the router with 5G backhaul:--- 10.0.0.36 ping statistics ---
9080 packets transmitted, 9080 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.494/3.844/1796.147/21.845 ms
--- 10.0.0.2 ping statistics ---
9083 packets transmitted, 9083 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.181/3.478/1953.779/21.589 msFrom the laptop (10.0.0.36) Ethernet-wired to the satellite through a switch with 5G backhaul:
Ping statistics for 10.0.0.1:
Packets: Sent = 10000, Received = 10000, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 282ms, Average = 0msPing statistics for 10.0.0.2:
Packets: Sent = 10000, Received = 10000, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 1ms, Average = 0msPing statistics for 10.0.0.33:
Packets: Sent = 10000, Received = 9883, Lost = 117 (1% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 17ms, Average = 0ms
I then connected the Ethernet backhaul cable:From the laptop (10.0.0.19) connected to the router through wifi with Ethernet backhaul:
-- 10.0.0.1 ping statistics ---
10000 packets transmitted, 9999 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.856/3.826/375.312/8.331 ms--- 10.0.0.2 ping statistics ---
10000 packets transmitted, 9491 packets received, 5.1% packet loss
round-trip min/avg/max/stddev = 1.011/4.137/375.312/9.731 ms--- 10.0.0.4 ping statistics ---
10000 packets transmitted, 9989 packets received, 0.1% packet loss
round-trip min/avg/max/stddev = 1.037/3.873/375.340/7.215 ms--- 10.0.0.33 ping statistics ---
10000 packets transmitted, 9524 packets received, 4.8% packet loss
round-trip min/avg/max/stddev = 1.038/4.202/504.643/11.575 msFrom the laptop (10.0.0.33) Ethernet-wired to the router with Ethernet backhaul:
--- 10.0.0.1 ping statistics ---
10000 packets transmitted, 9588 packets received, 4.1% packet loss
round-trip min/avg/max/stddev = 0.269/0.442/6.942/0.162 ms--- 10.0.0.2 ping statistics ---
10000 packets transmitted, 9486 packets received, 5.1% packet loss
round-trip min/avg/max/stddev = 0.242/0.362/1.081/0.053 ms--- 10.0.0.36 ping statistics ---
10000 packets transmitted, 9564 packets received, 4.4% packet loss
round-trip min/avg/max/stddev = 0.285/0.700/14.856/0.201 ms
- ekhalilMaster
I have wired backhaul so I thought to try to simulate your case, so I connected my laptop with a LAN cable to the router, disconnected the wifi network (did not disable wifi) in the laptop, started command prompt and typed the ping message, and guess what, I got packet loss exactly as you did, BUT.....
While trying to collect some date from Orbi I noticed that my speakers (usually connected to the satellite) started to blink indicating no network then I saw they got attached to the router, waited few minutes, I saw that my router rebooted, so I disconnected the laptop.
When I looked in the laptop IP configuration I noticed that I had it set up for a static IP address (192.168.10.101) outside the Orbi subnet and remembered that I did this setting last time when I was configuring a VoIP router for a friend. I really don't know how I managed to get LAN connectivity with this settings but it seems that somehow the wifi was sending the packets not the LAN port.
Of course the family at this point started to complain about loss of internet and I did not want to try again.
What I concluded from this short test -together with my previous experience with Orbi- is the following:
- Orbi LAN ports are very sensitive to the wired devices connected to it
- This "sensitivity" is more evident when you have wired backhaul
- I should be cautious with devices that have wireless capability when using them as wired devices in Orbi, they can cause loops and system instability.
At this point I'm not sure what made my Orbi unstable when I connected my laptop, is it the bad LAN port configuration, or the loops caused by combination of wired and wireless connections in the same device.
I will try again later and see if I can simulate this case again. Meanwhile I have the following questions:
- Do you have any static IP settings in your wired computers?
- Are you sure that wireless is completely disabled in those computers
- Does the same issue happen if you disconnect one of the computers (one at a time) and ping from the other computer to any other device (wired or wireless), repeat the the same with the other computer? It can be only one of the computers is causing this issue.
- Is Daisy Chain Disabled? you mentioned that but please make sure it's unticked.
Another test that I would like to do, at later stage, to avoid loops in backhaul, is to disable the wireless backhaul (with telnet commands) and see if this will do any improvement.
- MoooseLuminary
Thanks for looking into this!
I believe I have never set any static IP addresses, and I am sure the two laptops used for the test received theirs through DHCP.
I am certain macOS and Windows 10 on the two computers involved think wifi is disabled. I have also verified that they do not show up as wireless connected devices in the Orbi web interface.
I have done similar tests with these two particular devices turned off, it was packet loss from the Drobo to the Apple TV that originally alerted me that something was wrong.
I will try turning them off one at a time tomorrow and ping from the wifi laptop.
Daisy Chain is disabled. (Or, unticked, at least.)
I am surprised wireless backhaul cannot be easily disabled from the web interface.
- FURRYe38Guru - Experienced User
I would try doing a factory reset on the system and setup from scratch and run some test to see if the problem continues with ethernet backhaul connected with the satellite.
Is there any ethernet switches in between the router and satellite or is the satellite directly connected to the router?
No need to check the ISP service. Reason for mentioning this is that your devices are getting a 10.#.#.# IP address which means the Orbi detected a DHCP server in front of the Orbi which maybe using the same IP address as the Orbi does, 192.168.1.1 so the Orbi will change the default IP address for the LAN side to something else. In this case 10.
- Chuck_MMentor
Here's a really silly question:
How confident are you that your ethernet cable used for backhaul is configured and terminated properly?
Could the issue be something as simple as a defective cable connection?
- CrimpOnGuru - Experienced User
In a previous post, he mentioned substituting a new short CAT6 cable for the backhaul ethernet cable (which probably ran through the walls, over the ceiling, etc.). This situation defies belief, and he has done a TON of work to document it.
- MoooseLuminary
I have been trying to post an update since last night. It looks like posts are successfully submitted (success, click here to go to your post) but then they disappear. Looking around the forum interface I found a list of rejected posts, but none on the awaiting approval list.
The post is only text, 4262 characters, copied from a plain text editor. What am I missing?
- ekhalilMaster
Mooose wrote:
I have been trying to post an update since last night. It looks like posts are successfully submitted (success, click here to go to your post) but then they disappear. Looking around the forum interface I found a list of rejected posts, but none on the awaiting approval list.
The post is only text, 4262 characters, copied from a plain text editor. What am I missing?
Anything that looks like a phone number probably?
- FURRYe38Guru - Experienced User
Possible the Forum software is detecting your pastes incorrectly as spam thus deleteing the post. Try using Notepad++ to copy and paste from next time.
So if you connect the satellite in the same room via ethernet and let it settle in, then run some wired pings at the router and satellite while in the same room in this configuration, you see PLs here with a good quality CAT6 LAN cable? Even after a factory reset of the system and setup from scratch?
Download PingPlotter.com and run some tests by pinging the routers IP address with this program..
I'm wondering if your system maybe just bad and needs to be replaced. :smileyfrustrated:
Mooose wrote:
I have been trying to post an update since last night. It looks like posts are successfully submitted (success, click here to go to your post) but then they disappear. Looking around the forum interface I found a list of rejected posts, but none on the awaiting approval list.
The post is only text, 4262 characters, copied from a plain text editor. What am I missing?
- ekhalilMaster
Mooose Good troubleshooting and interesting observations. It's clear that the packet loss is directly related to the wired backhaul.
I'm now running the ongoing beta software (2.3.0.23) and I can see that NG has done quite good improvements on wired device handling. Would you want to test this SW and see if you will see improvements? If yes, please PM to ChristineT and ask to get the SW download links.