NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
djaesthetic
Dec 31, 2020Tutor
Disabling automatic subnet reconfig
I was wondering if there was a way (yet) to disable Orbi’s automatic IP reconfig if it detects a “conflicting” subnet? I understand the spirit of what it’s for, but frankly it’s a bugged feature (yes,...
- Jan 01, 2021
CrimpOn : Did some final testing and now confident in the conclusion.
What triggers the "reconfiguration" behavior appears to be whenever Orbi detects any other network device on the other side of it's Internet (WAN) port sharing the same subnet. It doesn't matter if there's an actual conflict or not -- simply it's existence. From a consumer support standpoint this is actually a pretty clever mechanism (though I wish they'd give us the option to disable it for various use cases).In my *personal* case - the issue was that I'd put the ports in their respective VLANs (10 for WAN, 20 for LAN) but left VLAN 1 in place. Regular (untagged) traffic was working just fine. My guess is that during a firmware update on those GS108Ev3 switches, it was sending out a broadcast across all configured VLANs, Orbi was seeing that broadcast on the Internet port, hence reconfiguration is triggered. I removed VLAN 1 from all ports and haven't been able to replicate the problem since.
As for your question about the use case for the two switches? This as a method to extend out multiple networks over a single cable. My current configuration looks like this:
VLAN 10 = WAN Traffic
VLAN 20 = LAN Traffic
-----BASEMENT
GS108Ev3 - Port 1: VLAN 10 Tagged, VLAN 20 Tagged
GS108Ev3 - Port 2: VLAN 10 Untagged
GS108Ev3 - Port 3-8: VLAN 20 UntaggedAT&T Gateway LAN plugged in to GS108Ev3 Port 2
UPSTAIRS OFFICE
GS108Ev3 - Port 1: VLAN 10 Tagged, VLAN 20 Tagged
GS108Ev3 - Port 2: VLAN 10 Untagged
GS108Ev3 - Port 3-8: VLAN 20 Untagged
Orbi Internet Port plugged in to GS108Ev3 Port 2
-----
Port 1 is the single physical cable running between the two switches. It will pass all traffic for either VLAN (LAN or WAN side) without either seeing one another as the traffic is "encapsulated" (isolated from each other). Port 2 on each side is where you plug in the WAN side of things. In the basement I have several runs from around the house plugged in to ports 3-8 (LAN). In the Office, I also have a bunch of devices plugged in to 3-8 (LAN). Two separate floors but they'll all end up in the LAN side.The notion that someone shouldn't be using managed switches in a network topology is a silly one, assuming the configuration is correct. In my particular case (and the fix to the original problem I posted about) turns out to simply be "don't let Orbi's Internet port see any traffic with a subnet that matches it's LAN side". Simple enough, makes a lot of sense. Once I understood what was triggering the reconfigurating, finding the root cause was simple.
(Extra thanks to schumaku for the sentence that led to the conclusion -- "Somehow the Orbi system does see any 192.168.x.x network on it's WAN/Internet port.")
djaesthetic
Dec 31, 2020Tutor
FURRYe38 : *WHICH* subnet is used should be irrelevant as long as there is no conflict. I also confirmed the issue existed both pre and post FW upgrade, so the upgrade wasn’t the cause.
I keep trying to drag the conversation back to the layer 2 switches as that’s the behavior that makes the least sense. They have zero L3 routing capability. A “conflict” with them is impossible unless one tried configuring their IP for the same as the router. The behavior is repeatable when kicking off a firmware update (on the switches). Their factory default IPs are 192.168.0.239 so even if during their upgrades they somehow momentarily defaulted back to that network, it wouldn’t overlap with any existing network. If I don’t understand the specific mechanism that’s causing the behavior, I can’t confidently prevent it from happening again at an arbitrary time.
(I’m a network architect so if you *do* understand what’s happening under the hood, please don’t hold back in explaining!)
I keep trying to drag the conversation back to the layer 2 switches as that’s the behavior that makes the least sense. They have zero L3 routing capability. A “conflict” with them is impossible unless one tried configuring their IP for the same as the router. The behavior is repeatable when kicking off a firmware update (on the switches). Their factory default IPs are 192.168.0.239 so even if during their upgrades they somehow momentarily defaulted back to that network, it wouldn’t overlap with any existing network. If I don’t understand the specific mechanism that’s causing the behavior, I can’t confidently prevent it from happening again at an arbitrary time.
(I’m a network architect so if you *do* understand what’s happening under the hood, please don’t hold back in explaining!)
djaesthetic
Dec 31, 2020Tutor
FURRYe38 : It’s repeatable that a firmware update on the switches triggers the issue. That already proves they’re one of the causes. I don’t need to remove them to prove what we’ve already established. That doesn’t answer the “why”.
I need the Orbi network to remain 192.168.1.0/24. There are no other layer 3 devices on the network between that and the ATT GW (currently on 172.16.0.0/24).
My DHCP pool is already reduced to .150 - .190 with all static addresses outside of those.
I need the Orbi network to remain 192.168.1.0/24. There are no other layer 3 devices on the network between that and the ATT GW (currently on 172.16.0.0/24).
My DHCP pool is already reduced to .150 - .190 with all static addresses outside of those.
- FURRYe38Dec 31, 2020Guru - Experienced User
I would ask you to start a post thread here and post about this in this switch forum:
If this is being triggered by switches, then there maybe something on the switches that needs to be looked at and would not be a Orbi problem. Seems like someting with these managed switchs is causing Orbi to trigger the subnet config incorrectly. The one managed switch I have and use doesn't do this, however it's a D-Link DGS-1100 and it's set for a static IP address of 192.168.0.11 which is out side of the DHCP IP address pool on my router.
Good Luck.
- djaestheticDec 31, 2020TutorFURRYe38 : Not a bad idea! Perhaps someone over there may be familiar with what happens under the hood when a firmware update is initiated.
Irony be damned, I almost bought a pair of DGS-1100 instead but opted for the GS108Ev3 simply because “both were Netgear”.
I *did* want to reiterate the IPs in play:
Orbi: 192.168.1.1
SW1: 192.168.1.2
SW2: 192.168.1.3
Orbi DHCP: 192.168.1.150 - 192.168.1.190
Static Devices: All .200 and above
The *default* IP of the switches out of the box is 192.168.0.239.
There is no overlap anywhere. - FURRYe38Dec 31, 2020Guru - Experienced User
Ya, seems odd that switches are causing this. Even when out of the box, .0.### shouldn't trigger Orbi to change subnets if Orbi is on 1.1. Like I said, I haven't seen this before so seems like something on the switch side does something that Orbi sees falsely. Seen odd things with Orbi and managed switches. Would be one test to try a non managed switch in place to see if the same thing happens. I presume you shouldn't see this.
Hopefully someone on the switch side can point you in the right direction.
- djaestheticDec 31, 2020TutorFURRYe38 : Wish I could use unmanaged switches. Actually swapped *OUT* a pair in place for these managed ones.
I have a single Cat6 run between two floors of my house. My internet comes in to the basement. About 6 other physical runs terminate in the same location. My router is in my office one floor up. I needed a way to connect all of those physical LAN connections *AND* the internet WAN connection over a single cable. The solution was to put in two managed switches. The single Cat6 run between floors is a trunk passing *ALL* VLANs on Int1. Int2 (on both sides) is the WAN VLAN (where the ATT GW plugs in on one side and the Orbi “Internet” port plugs in on the other). Int3-8 on both sides are dedicated to LAN. It’d work beautifully if it weren’t for this Orbi “feature”.
(...I also have 3 floors to my house, so moving the router on the basement would be too detrimental to Wi-Fi on the top floor, so that’s out...) - FURRYe38Dec 31, 2020Guru - Experienced User
Mine just about opposite of your house. Two story with basement. ISP modem and router on 2nd floor upstairs. All in electronic closet. Then I have patch panel for CAT6A to wall outlets on 2nd and some on 1st floor. I have 1 non mananged switch segment connected to the back of the RBR, NG GS-108 that connects my DISH Hopper system, ATT microcell in the living room and 1 line that goes down to the living room on 1st floor where 1 non managed switch and microcell is, then this continued line that goes to the back far office room down the hallway where the original office and networking was a few years back. Here there is a wall jack we put in the runs from this room down to the basement area where a tennant lives and 1 non managed switch and last RBS is connected on this line. The living room and basement switches use NG GS-105s. Basement has two xbox consoles, power switch and a PC. All this works with no VLANs or managed switches.
- djaestheticDec 31, 2020TutorFURRYe38 : “ All this works with no VLANs or managed switches.”
...well, YEAH. Only a bunch of other hardware and cable runs instead. Heh That’s exactly what I’m avoiding here. I manage enterprise networks for a living. A couple of managed switches to pass a pair of VLANs should not be problematic in the slightest. This is basic Networking 102 stuff... - FURRYe38Dec 31, 2020Guru - Experienced User
Hopefully you can figure out whats going on with the switches. Possible just a configuration issue.
Good Luck.
- CrimpOnDec 31, 2020Guru - Experienced User
djaesthetic wrote:
FURRYe38: Wish I could use unmanaged switches. Actually swapped *OUT* a pair in place for these managed ones.
I have a single Cat6 run between two floors of my house. My internet comes in to the basement. About 6 other physical runs terminate in the same location. My router is in my office one floor up. I needed a way to connect all of those physical LAN connections *AND* the internet WAN connection over a single cable. The solution was to put in two managed switches. The single Cat6 run between floors is a trunk passing *ALL* VLANs on Int1. Int2 (on both sides) is the WAN VLAN (where the ATT GW plugs in on one side and the Orbi “Internet” port plugs in on the other). Int3-8 on both sides are dedicated to LAN. It’d work beautifully if it weren’t for this Orbi “feature”.What a puzzle. Sorry to have no inside scoop on how Orbi functions with regard to IP address conflicts. There are questions all the time from people who have only a single ethernet cable to the Orbi location and buying a couple of switches would be vastly less costly than tearing open the walls to install more cable.
I do wonder, however, if part of the first problem was the Orbi DHCP subnet mask. The Orbi subnet mask was changed from 255.255.255.0, correct? Having "no conflict" in the actual DHCP ranges may be less significant than an "implied conflict".
On the switch issue, can the problem be repeated by power cycling the switch? Both the GW and Orbi are responding to DHCP, correct requests, correct? Which one will the switch get an IP from? I can imagine the Orbi observing, "My internet gateway is 172.x.x.x, and 'oh, my' there's a device on my LAN which is in the same subnet. How about hard coding static IP's in the two switches?
Or... sorry I don't understand.
- schumakuDec 31, 2020Guru - Experienced User
"Disabling automatic subnet reconfig" - you can't disable this wonderful commodity ***. Just ensure no 192.x.x.x network ever shows up on the Orbi WAN port - and nothing strange will happen.
For anything else - follow my reply on you other thread -> GS108Ev3 firmware update triggering Orbi subnet reconfiguration - with your qualification a snap 8-). Up front: It's not a switch issue.
- djaestheticDec 31, 2020Tutor
CrimpOn : Appreciate the response!
I'd considered the DHCP idea, except Orbi doesn't allow you to define a proper subnet for DHCP. Instead they just ask for a range (i.e. there's no subnet mask defined re: DHCP).
Oddly enough, the issue is NOT repeated by power cycling the switch. After seeing the firmware update issue, I'd assumed a power cycle would cause the same behavior -- but it doesn't.
The GW has a DHCP range set for two IPs (172.16.0.60-61) solely for performing a WAN IP passthrough back to Orbi's WAN interface (public WAN IP). Orbi has a DHCP range set for 192.168.1.150 - 192.168.1.190.
Both switches have hardcoded IP addresses of 192.168.1.2 and 192.168.1.3 respectively.
I think another poster may have dropped a clue in the other thread regarding specifically would be triggering this (i.e. Orbi's *WAN* port seeing an IP address that matches the subnet of it's LAN configuration), so going to focus the conversation there. I'm seriously thinking VLAN 1 is my likely culprit. Just trying to figure out how to validate it before sharing the final cause in these threads for the next poor !@#$!%^! who stumbles across the issue.
- CrimpOnDec 31, 2020Guru - Experienced User
So this is totally stable now, every after power cycling of all components? Two switches for about $120 beats the heck out of installing another ethernet cable.
- djaestheticDec 31, 2020TutorCrimpOn : Seems it. I’ve bounced both switches, the AT&T gateway, and the Orbi - **ALL** at least twice and no issues. I wish I could test a firmware update on those switches (as that reliably triggered the problem before) just to be sure — though I really do feel the removal of VLAN 1 from all ports was the likely culprit (once I understood what specifically was causing Orbi to do this in the first place).
- FURRYe38Dec 31, 2020Guru - Experienced User
Just curious, can you downgrade one of the switches, then re-upgrade to see if this triggers anything? Just an idea...
- CrimpOnDec 31, 2020Guru - Experienced User
djaesthetic wrote:
CrimpOn: Seems it. I’ve bounced both switches, the AT&T gateway, and the Orbi - **ALL** at least twice and no issues. I wish I could test a firmware update on those switches (as that reliably triggered the problem before) just to be sure — though I really do feel the removal of VLAN 1 from all ports was the likely culprit (once I understood what specifically was causing Orbi to do this in the first place).On a scale of 1 to 10, how difficult would you describe the process of setting up these switches?
There have been many posts from people who discover that their ISP connection feeds to a wiring hub in the basement with ethernet cables running all over the house, but only one to each location. Baements are a horrible location for WiFi routers, especially if the only thing "down there" is the wire center. If not too difficult to implement, this would be of help to such people.
- CrimpOnJan 01, 2021Guru - Experienced User
Looking further, this is SO COOL. A pair of 5-port TP-Link managed switches is only $50. (or, a 5-port at the Orbi location and a larger switch at the wiring hub would be more useful). Still WAY less than the cost of placing an ethernet cable in a finished structure.
(Waiting for the shoe to drop....) There have been MANY posts expressing frustration with connecting Orbi satellites to the router through a switch. Does this network design extend to connecting satellites over ethernet, or are they connected using the default WiFi backhaul?
The common wisdom is to avoid "managed switches", which clearly includes these.
- schumakuJan 01, 2021Guru - Experienced User
CrimpOn wrote:Looking further, this is SO COOL.
Welcome to the world of networking in 2021 my friend!
CrimpOn wrote:A pair of 5-port TP-Link managed switches is only $50.
Never mind, we're on the Netgear community here. Guess you talk of the Netgear Smart Managed Plus like TP-Link Easy Smart/Unmanaged Pro Switches which have very much in common.
All these devices are not built on managed cores, these are much more configureable switch cores combined with a tiny 8051 class microcontroller.
Two major differences apply:
- There is no STP (or RSTP, or MST) - this is avoiding the issues people have with Orbi and Web/Smart managed and true managed switches.
- There is no management VLAN. In DHCP mode the switch will pick-up an address from any VLAN (this can be the cable modem if it's a plain IP one). Similar, the management IP can be reached on any VLAN. Few models support the ability to configure a management IP to "restrict" the access.
The second point does (in my opinion) disqualify these switches for exposing to the wild Internet.
CrimpOn wrote:There have been MANY posts expressing frustration with connecting Orbi satellites to the router through a switch.
Root cause is always the STP on by default on almost all Web/Smart managed and true managed switches, the non-compatible STP default config, and Netgear's lack of documentation on the Orbi STP design.
CrimpOn wrote:Does this network design extend to connecting satellites over ethernet, or are they connected using the default WiFi backhaul?
Wired backhaul of course.
CrimpOn wrote:The common wisdom is to avoid "managed switches", which clearly includes these.
This wisdom is OK to keep people away doing things requiring more than pluging a cable and start shouting about the bad Orbi or bad switch maker instead. 8-)
- djaestheticJan 01, 2021TutorCrimpOn : Actually setting it up was pretty simple, minus the obvious problem that started this thread to begin with. This morning I realized a way I could test my theory as to what was causing it. If I’m correct, I’ll update all threads with the final root cause and solution, plus I’ll go ahead and throw out how to do the switch design for users wanting to “move” where their routers are.
- djaestheticJan 01, 2021Tutor
CrimpOn : Did some final testing and now confident in the conclusion.
What triggers the "reconfiguration" behavior appears to be whenever Orbi detects any other network device on the other side of it's Internet (WAN) port sharing the same subnet. It doesn't matter if there's an actual conflict or not -- simply it's existence. From a consumer support standpoint this is actually a pretty clever mechanism (though I wish they'd give us the option to disable it for various use cases).In my *personal* case - the issue was that I'd put the ports in their respective VLANs (10 for WAN, 20 for LAN) but left VLAN 1 in place. Regular (untagged) traffic was working just fine. My guess is that during a firmware update on those GS108Ev3 switches, it was sending out a broadcast across all configured VLANs, Orbi was seeing that broadcast on the Internet port, hence reconfiguration is triggered. I removed VLAN 1 from all ports and haven't been able to replicate the problem since.
As for your question about the use case for the two switches? This as a method to extend out multiple networks over a single cable. My current configuration looks like this:
VLAN 10 = WAN Traffic
VLAN 20 = LAN Traffic
-----BASEMENT
GS108Ev3 - Port 1: VLAN 10 Tagged, VLAN 20 Tagged
GS108Ev3 - Port 2: VLAN 10 Untagged
GS108Ev3 - Port 3-8: VLAN 20 UntaggedAT&T Gateway LAN plugged in to GS108Ev3 Port 2
UPSTAIRS OFFICE
GS108Ev3 - Port 1: VLAN 10 Tagged, VLAN 20 Tagged
GS108Ev3 - Port 2: VLAN 10 Untagged
GS108Ev3 - Port 3-8: VLAN 20 Untagged
Orbi Internet Port plugged in to GS108Ev3 Port 2
-----
Port 1 is the single physical cable running between the two switches. It will pass all traffic for either VLAN (LAN or WAN side) without either seeing one another as the traffic is "encapsulated" (isolated from each other). Port 2 on each side is where you plug in the WAN side of things. In the basement I have several runs from around the house plugged in to ports 3-8 (LAN). In the Office, I also have a bunch of devices plugged in to 3-8 (LAN). Two separate floors but they'll all end up in the LAN side.The notion that someone shouldn't be using managed switches in a network topology is a silly one, assuming the configuration is correct. In my particular case (and the fix to the original problem I posted about) turns out to simply be "don't let Orbi's Internet port see any traffic with a subnet that matches it's LAN side". Simple enough, makes a lot of sense. Once I understood what was triggering the reconfigurating, finding the root cause was simple.
(Extra thanks to schumaku for the sentence that led to the conclusion -- "Somehow the Orbi system does see any 192.168.x.x network on it's WAN/Internet port.")
Related Content
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!