I have just started using OpenWRT on a Netgear R6700v2 (MIPS/Mediatek). I am interested in using the device for a client bridge. I am not a fan of WDS since it hurts wireless bandwidth and so am not considering that topology.
I see this tutorial on this sort of setup and have a few questions:
The tutorial says: "In some cases, the wireless drivers used in OpenWrt do not support “Layer 2” bridging in client mode" - in what cases CAN OpenWrt support this? Is a client bridge configuration possible in OpenWrt without relayd? Can I not set up one radio as a client and just bridge it to the LAN?
The tutorial describes how to allow LAN traffic on router2 to be relayed, but not wlan traffic. With or without relayd, is it possible in OpenWRT to use one radio as the client radio, and the other to allow connections to?
Thanks.
[EDIT] The discussion got (as I figured it would) side tracked to a discussion of the technical aspects and merits of WDS. That has subsequently been tried and isn't working for my setup. Please, if you have anything to offer on the actual questions asked above, it would be appreciated.
There's no bandwidth issues if you use a radio just for your WDS. Those issues only play if you'd be using the same radio as an access point simultaneously.
My WDS setup is on the 2,4 GHz channel with the actual access points on 5 GHz. Would that be an option for you?
There's no bandwidth issues if you use a radio just for your WDS.
When I dedicate a radio just to router-to-router communication, then a) I lose that radio on both sides, and b) the routers can only talk at that link's speed. I wouldn't cry over losing 2.4GHz, but then anything on the secondary router can only talk to all the devices on the primary router at the speed of tghe 2.4GHz link. WDS is also problematic at times between chipsets. I'd much rather go through the initial pain to get a client bridge working properly.
WDS (i.e. 4-address mode) is the proper way to set up a client bridge. Using relayd instead doesn't offer any advantage, and would only make sense if one of the devices doesn't support WDS.
On the access point side, there is no need to dedicate a radio for this. Normal clients can still connect as usual to an access point with WDS (i.e. configured with mode ap and wds 1, or as "Access Point (WDS)" in LuCI).
If you want the client to also work as an access point on the same radio, then that will affect performance. But this is unavoidable in that case, and has nothing to do with WDS.
Router 1 - "primary" (has the external WAN connection)
Router 2 - "extender" (no WAN)
5GHz network SSID RadioActive5
Router 2 connected to Router 1 on RadioActive5 using WDS
It is my understanding with given the above that I have two choices with WDS:
I can use WDS repeater modes, in which I can have client devices also connect to RadioActive5 but when they do I cannot control which actual router any client device connects to. If a client device happens to be closer to Router 2 and connects to RadioActive5, then it will likely actually make its connection to Router 2. And then any packets that device sends destined for the internet would go over RadioActive5 from the device to Router 2 and then from Router 2 to Router 1.
I can use WDS AP/Client modes where no client device can connect to RadioActive5 and where that radio is dedicated solely to router-to-router communications
There is no specific WDS repeater mode in OpenWrt. The WDS setting in OpenWrt applies to any access point or client interface (configured by the option wds on a wireless interface, or by selecting a mode of "Access Point (WDS)" or "Client (WDS)" in LuCi). What that does is to enable 4-address mode, which is required to be able to bridge a wireless client to a wired network without using hacks like relayd.
Enabling WDS for the RadioActive5 interface on device 1 won't have any direct effect. All clients will still be able to connect to that wireless network as before. The only thing this does is to allow WDS clients to connect.
So, now you can configure the client interface with WDS on device 2, and create a bridge containing this interface. This still won't affect any other devices that are connected to device 1, which will continue to work normally.
Optionally, you could also add another access point interface (with or without WDS) on the same radio on device 2, which would make it a repeater (with the associated performance reduction). But this is not required. If you only want to use device 2 as a client bridge for wired devices, you can leave it at the client interface.
You're mixing software and hardware here. Not just OpenWrt and Tomato but also Mediatek (works, my WDS is Mediatek on both sides) and Broadcom (old proprietary blob).
I don't think 4addr works (well) on anything except Mediatek and Atheros. So in your case you might need relayd. And 99% of the WDS setups with OpenWrt are with OpenWrt on both sides.
Ok, so I've spent some time with relayd and here are the answers to my original questions (in case anyone comes along grepping for this later)
Q: Is a client bridge possible without relayd? Answer: Only with WDS. Some firmwares (notably DD-WRT) have a client bridge mode hacked in but it is essentially doing internally what relayd is doing. The only client bridge options in OpenWRT appear to be WDS and relayd.
Q: Can the second radio be added in to the bridge? Answer: It sure can. The tutorial even briefly touches on it. Once done the tutorial, you can bring up the second radio as per normal and it will go through relayd.
It's pretty slick and, so far, is rock solid stable.
Thanks for the post. It's exactly what I needed to accomplish. I brought up the second radio attached to the [lan] network and my client picked up an IP and was able to communicate with the wireless network the first radio was connected to.