• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Question Losing Networking when going from WiFi AP to Router WiFi

tekz

Junior Member
Oct 15, 2019
12
1
11
Hi all, I made a hasty networking diagram to describe this issue below:



Appreciate any help, I'm out of ideas :/
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
What's router's LAN IP? 192.168.1.254 or 192.168.1.1? Why 2 IP addresses?

A bit confusing here.
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
So you mean router LAN IP is 192.168.1.1 and the DHCP server on router 1 is giving IP addresses ranging from 192.168.1.1 to 192.168.1.254 , and subnet mask is /24?

Anyway, I assume you are using same SSID for router 1 and router 2? In that case, the roaming for wifi client might not work properly since its not a modern mesh system.

==

Oops, you are using different SSID. :rolleyes:

@JackMDS might have a better answer.

==

No idea why you have to do MAC to IP binding or unbinding? on LAN or WAN side? What's the purpose of doing that? And how did you do that? What's router 1's brand/model?

Besides, why does the whole thing has anything to do with VLAN? I'm not good at VLAN though.
 
Last edited:

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
OK, since you have 2 router/AP in the same IP range 192.168.1.1 to 192.168.1.254, and signal can't cover both rooms,

1. Set router in room 1's DHCP server to give out IP from 192.168.1.10 to 192.168.1.120 and

2. enable AP's DHCP server in room 2 to give out IP from 192.168.130 to 192.168.1.240

so IP range won't conflict. See if the setup works.
 

VirtualLarry

No Lifer
Aug 25, 2001
52,241
7,058
126
Is Router 2 connected to the powerline adapter, via it's WAN port, or LAN port? I'm thinking that it should be connected via LAN port, and honestly, I think it's DHCP server should be disabled. I think that's botching things up here.

I guess I'm a little confused with your diagram. Router 2, is shown as "DHCP: Disabled", but yet, "DNS: PI-Hole (IP)"? If DHCP is truely disabled, it's not going to be handing out DNS server IPs to clients with PI-Hole (IP). Do you mean that the router's own DNS is set to the PI-Hole? If so, consider setting that to Router 1 as the DNS server for Router 2. (Router's OWN DNS server, not the DNS server handed out via DHCP to clients, right?)
 

tekz

Junior Member
Oct 15, 2019
12
1
11
Hi all, apologies, in my haste I've led to confusion, didn't have a lot of time to post so threw up that diagram, to answer some of your questions:

MAC address binding it for management hosting of different things on different clients, it's not essential and I switched it off while testing for a while.

VLAN management can be ignored, I was trying to give a scope of the network as I have no idea why this is happening but I don't think it's to do with VLAN. This part of the network is all on the same VLAN anyway and the router is set to allow all non tagged traffic on to this subnet (even if it doesn't support VLAN tagging).

Main Room:
Router = Draytek Vigor 2860
IP address is 192.168.1.254, the LAN range denoted by CIDR is 192.168.1.254/24 (so subnet mask is 255.255.255.0).
DHCP is on (but I have IP to MAC binding enabled so clients keep their IP's. I have some on static IP outside of the DHCP pool, I.E. the Pi-Hole but this is probably not important.
DNS for DHCP clients is set to the PI-Hole.
Due to VLAN and VPN I have to use this DHCP server.

Second Room:
Wireless AP (It doesn't perform NAT, it's not a router).
The second device is a TP-Link WiFi extender, it has two modes, wireless extender (forwards WiFi packets) or WiFi access point (ethernet bridge). It is set up as a WiFi access point. Connected via ethernet to port 1 on the router (VLAN 1/Same subnet) through the powerline adaptors and has a static IP set up: 192.168.1.253, subnet mask 255.255.255.0, DNS set to Pi-Hole (for it's own traffic).
DHCP server is off on this device, so clients obtain DHCP information from the Router 1. This does work, but like I said, when I switch clients from the Wireless AP, to Router 1 SSID (they have different SID's, on different channels), the clients have no networking, I have to reboot them or reboot the whole network.

I guess I'm a little confused with your diagram. Router 2, is shown as "DHCP: Disabled", but yet, "DNS: PI-Hole (IP)"? If DHCP is truely disabled, it's not going to be handing out DNS server IPs to clients with PI-Hole (IP). Do you mean that the router's own DNS is set to the PI-Hole? If so, consider setting that to Router 1 as the DNS server for Router 2. (Router's OWN DNS server, not the DNS server handed out via DHCP to clients, right?)
I'm not sure I see the point in this VirtualLarry, the problem is not while connected to the Wireless AP but when going from the AP SSID to the Router SSID, this would also negate the DNS filtering from the PI-Hole and lead to a DNS leak, and I don't think this is a DNS issue as IP's can't be resolved by lookups either when the clients have no networking.

To outline the problem:
1. I connect my laptop or mobile to the router WiFi, everything is working.
2. I walk in to the other room, my device connects to the AP wifi, everything is working, my IP is the same as on the previous WiFi due to IP to MAC binding etc, DNS is routing properly through the PI-Hole, etc. Everything working how it's supposed to.
3. I walk back in to the main room with the Router and connect to it's WiFi again. It connects, I have the same IP as before, but no traffic, no response to pings internally using hostame or IP, no internet. If I'm on a PC and perform an ipconfig/release/renew it does nothing, it waits and waits. It's like it's connected but not connected. If I remember correctly while testing on Sunday night, on a few occasions I could ping the Router while the clients in this state but nothing else, most of the time I couldn't ping anything. Oddly when I reboot the client, the issue goes away, until I connect to the Wireless AP again, rinse and repeat. I think this might be some layer 2 issue or something but I'm out of my depth.
 
Last edited:

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
If router IP is 192.168.1.254 with /24 denotation then the network is 192.168.1.0, the correction denotation is 192.168.1.0/24 not 192.168.1.254/24. And what is this IP address 192.168.1.1?

In room 1 there are 2 devices capable running DHCP servers - Draytek Vigor 2860 & Pi-Hole. Did you enabled both DHCP servers?

Or probably 3, did you check NAS server?

Still confused.

 
Last edited:
  • Like
Reactions: VirtualLarry

VirtualLarry

No Lifer
Aug 25, 2001
52,241
7,058
126
I think this might be some layer 2 issue or something but I'm out of my depth.
I was wondering about broadcast domains, with the powerline in the mix, and ARP issues (ARP caching / proxying by the router or powerline).

Possibly someone who knows more than I (not hard) can shed some more light on possible issues here.
 

tekz

Junior Member
Oct 15, 2019
12
1
11
If router IP is 192.168.1.254 with /24 denotation then the network is 192.168.1.0, the correction denotation is 192.168.1.0/24 not 192.168.1.254/24. And what is this IP address 192.168.1.1?

In room 1 there are 2 devices capable running DHCP servers - Draytek Vigor 2860 & Pi-Hole. Did you enabled both DHCP servers?

Or probably 3, did you check NAS server?

Still confused.

Apologies this isn't my field, you are correct with: 192.168.1.0/24, just checked it in router/google CIDR calculator, 192.168.1.1 is nothing. DHCP is disabled on the PI, it's disabled on the TP-Link AP, only DHCP server is on the Draytek, I need to use the Drayteks one for other reason (has some applications you need to use that DHCP server for).

I was wondering about broadcast domains, with the powerline in the mix, and ARP issues (ARP caching / proxying by the router or powerline).

Possibly someone who knows more than I (not hard) can shed some more light on possible issues here.
This is where I was thinking but my networking knowledge gets even more flakey here, I don't know enough about powerlines and how they function, or ARP tables and how they renew or anything that could be causing this, it seems simple enough to use an AP over a powerline to a router, I'm not asking for much, it's what they were designed for so don't know why I'm getting this issue or what to do to try to find out. Odd thing is the clients get an IP address when they aren't functioning. In theory broadcast domain should be ok as it's all on the same LAN?
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
Interesting.

Like what VL said, it's very possible an ARP table issue, but I'm not good at that part either.
 

VirtualLarry

No Lifer
Aug 25, 2001
52,241
7,058
126
The AP and the router have different SSIDs, right?

Maybe this would be a useful experiment, temp take the powerline out of the equation. Wire the AP to the router, no change in settings, just run a (long, if needed) Cat5e cable between router an AP, and try to transition your client devices back and forth between SSIDs. If they work, then chances are it's a proxy-ARP issue with the powerline bridge setup.
 

tekz

Junior Member
Oct 15, 2019
12
1
11
That's a good suggestion, I'll try that tomorrow and let you know. I'll try to do a bit of ARP googling, see if anything comes up that might help too.
Thanks both.
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
What if you disable router's DHCP and enable Pi-Hole's DHCP server?

Restart everything after reconfiguration.
 

tekz

Junior Member
Oct 15, 2019
12
1
11
I'll try that too for testing, may be a bug with the routers DHCP server. Although the clients are technically bound to their IP address, I'll turn this off too and let you know.
 

VirtualLarry

No Lifer
Aug 25, 2001
52,241
7,058
126
Maybe the problem is that the default IP gateway router (Router 1) doesn't have a 192.168.x.1 IP, but a .254 IP, and one of the two active DHCP servers (?) is handing out IPs, but not default gateway IPs? Kind of a long shot, but I've seen weirdness like that before.
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
default gateway and static IPs (MAC address binding) can be set in Pi-Hole DHCP Server config.

 
  • Like
Reactions: tekz

tekz

Junior Member
Oct 15, 2019
12
1
11
@mxnerd I could try this, would reduce load on the router when the pi has plenty of resources, only problem is I reboot the PI often when playing with stuff (this isn't a problem with DNS as router takes over temporarily with ISP DHCP DNS). I don't really want this bring down the network. Also I want to find the root of the problem rather than a workaround if it's something to do with DHCP if possible.

@VirtualLarry could be a weird gateway bug, but remember the router 1 SSID works fine, only when I connect to the second network then back to the first the issue arises, I think you were more on track with ARP, as the first router does have a 192.168.1.254 static IP and there is only one DHCP server not two.

I guess eliminating the EOP devices night tell us if it's an ARP issue, could be an ARP issue with the router also. I'll do this first, then try using the other DHCP server on the PI.

Setting up a static route temporarily might also help. I'll have to pull out wireshark to scan WiFi and port mirroring I suspect at some point but not entirely sure what I'm looking for yet. :(
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
You can make DHCP lease time as long as possible. That way you don't have to worry too much about rebooting Pi-Hole.
 

tekz

Junior Member
Oct 15, 2019
12
1
11
Well it seems I was wrong when I said the VLAN wasn't important. I tinkered around for hours, tried everything suggested plus more, nothing worked, about to give up then thought might as well turn off the VLAN's and see what happens, my network was butchered at this point anyway. To my surprise it started working. Set my network up again and it's still working! Now it's good to have it honed in on something but I really don't know why this is occuring when the VLAN's are enabled, sounds like a software bug or could some setting cause this behaviour? Would like to resolve it as the VLAN's are part of my network security and would like to keep them.
 

mxnerd

Diamond Member
Jul 6, 2007
6,194
900
126
VLAN of course is going to affect the network since it creates a virtual LAN - a separate broadcast domain.

But I can't help here.
 
Last edited:

ch33zw1z

Lifer
Nov 4, 2004
34,314
13,229
146
What is the VLAN configuration?

do all devices support VLAN's?

unmanged switches won't pass tagged VLAN traffic...and SOHO gear can be iffy on VLAN anything.
 

tekz

Junior Member
Oct 15, 2019
12
1
11
Everything in question is on VLAN 1, all ports on the draytek router and wifi are on vlan 1 and subnet 1 is broadcast to all physical ports. It is port based no VLAN tagging is done, and untagged packets are permitted to VLAN1 (It's ticked by default so left it on but shouldn't apply). VLAN 2 (different subnet) is for one specific port which is on both VLAN 1 and 2 for a specific server. VLAN 3 is isolated and solely for a guest wifi network.
None of this should matter really as the whole network segment in question is the same, subnet 1, VLAN 1 and port based VLAN at that. Both the main draytek router WiFi, AP WiFI are on the same network/VLAN.
Even if they weren't and the AP was on VLAN 2 for example, inter-lan routing is enabled for subnet 1 and 2.

I did think this could be an incompatibility issue with VLAN traffic but the problem occurs when going back to the Dratek WiFi, which is the Dratek device itself, responsible for managing VLAN traffic and the WiFi, and the WiFi sits on VLAN 1 by an internal switch I guess, so it should work?
 

ch33zw1z

Lifer
Nov 4, 2004
34,314
13,229
146
Is the Dratek running latest firmware? The config as stated should generally work ok, but sometimes SOHO devices aren't that good at their stated functions lol.
 

ASK THE COMMUNITY