Client Roaming

TheDevopsGuy

Posts: 681   +195
Whats the point of disconnecting a client if hes just gonna connect to the old ap even when the neighbor ap has stronger signals...


meh.

/rant
 
Need to clear auto connect to allow manual choice of the SSID
Its a mesh network running on same channels one ssid and similar paramters except the singals and noise levels so thats not really an option jo.

Im probably gonna try to further attenuate the signals to force lower signals at longer distances to encourage roaming.
 
Selection is NOT based upon signal strength but user's choice.

If you "forget" all existing networks that are visible, then you will be able to choose without the existing bias.
 
Selection is NOT based upon signal strength but user's choice.

If you "forget" all existing networks that are visible, then you will be able to choose without the existing bias.

I honestly don't have a clue how that would be doable in the current topology jobeard. The scope is that I want roaming without any connection interruptions. The client when connecting to the SSID should have knowledge of all the 3 MAC BSSIDs in the mesh network and their signal levels. All the AP's are listed under one SSID, I cannot simply select one from another. All the selection of which mesh node to use is being controlled by the client preference.

All im in control is the distance between the client and the mesh node which improves parameters to force the client to connect to the best node.
 
To further explain myself the issue isn't exactly that the network isn't functioning properly. All is well, its only the tests I want to conduct are based on luck rather than having full control over them.
 
My comment is basically non-orbit wifi.
  • Taking CP->Network & Interenet->Manage Wireless
gives a list of ALL networks the PC has had a wifi connection with. Forget each and every one.

You then get to choose your SSID for the next connection.

For your Orbit, the SSID is common and you don't get to select which one of the nodes you're accessing. See this discussion:
https://community.linksys.com/t5/Ve...ng-to-the-closest-strongest-node/td-p/1195494

upload_2019-4-22_7-38-38.png
I did try something similar to what they were talking about. If the signal reaches -66 or higher for longer than 2 seconds the client should be forced to rescan the network. Unfortunately since all the AP's are quite next to each other their signals are very strong. This current Issue that im facing probably will be solved in the next scenario where im going to be attenuating the power of all the AP's in half.
 
That's a GREAT feature and interface :)

If only it works always. It's not working with all the devices either because the signal is not matching or a bug. Idk if messing with the authentication button would fix things as it would block the client from reassosiating with the current ap he just disconnected their features with mesh are quite wierd. I'll keep on tinkering I guess.
 
The object of the mesh is a reliable connection (by any means). I would not obsess with always connecting to the absolute closest station, as that will not ensure the best performance anyway. I hope you did your homework on the mesh technology as every additional station reduces the bandwidth to all clients due to the "chatty" behavior between stations.
 
The object of the mesh is a reliable connection (by any means). I would not obsess with always connecting to the absolute closest station, as that will not ensure the best performance anyway. I hope you did your homework on the mesh technology as every additional station reduces the bandwidth to all clients due to the "chatty" behavior between stations.

Ye, unfortunately mesh on one wireless band is heavily bottlenecked. I know it can be mitigated by dual band where you would pass mesh data from the 5GHz band and client data through the 2.4GHz. In all I want to observe the performance of the mesh network per hop, the degradation level and its sustainability. From just prior tests I can already see 50% network performance decreasing per hop, quite a big drop tbh.
 
Have a look at the SSID ORBI01 here

It's on channel 4 for both units. A better choice would have been to move one
of them to another channel, say 11. This would allow the client to pick the stronger signal.
 
Have a look at the SSID ORBI01 here

It's on channel 4 for both units. A better choice would have been to move one
of them to another channel, say 11. This would allow the client to pick the stronger signal.

Mesh kindoff relies on having the same channel/medium to communicate on, that would undoubtedly break the mesh bond between the orbi setup. I believe there are some mesh systems which you can setup with different channels but most of them don't.

Unfortunately one of the biggest limitations of the mesh networks is too much branching. A standard was created for a purpose, stick to it we have far too many corporations developing their own standards which in the end provide different results.
 
Mesh kindoff relies on having the same channel/medium to communicate on, that would undoubtedly break the mesh bond between the orbi setup.
OUCH! Then I can do better with two wifi router connected with some EoP device and manage the channels independently -- at a competitive price too!
 
OUCH! Then I can do better with two wifi router connected with some EoP device and manage the channels independently -- at a competitive price too!
Ye thats definetly part of the discussion jo. I don't know why ISP's locally are trying to shove wireless mesh systems to home users. You lose too much performance, you lose most mesh benefits in a small network and it's even more expensive to deploy or purchase as a package.

For home use stick to standard 802.11.

If and only if in the case where you have a SAN system internally or require low bandwidth RTP sesions then mesh is the system for you.

Companies and public infrustructures would benefit more from such systems if implemented correctly imho.
 
Mesh certainly has the Opportunity to do this right if:
  • the back channel (eg 5ghz) is separated from the client (2.4ghz)
  • each unit could be configured with a unique channel.
How the designer(s) could opt for loading the client channel with feedback noise is mindless imo.
 
Hmm now this is interesting.

I'm not certain 100% but throughout my testing on my linux pc while roaming through the mesh network it is taking too long, 6-10 pkts dropped to roam to the next AP. Thats Long!

Proceed to use my windows laptop in the same network.

No packets dropped at all.

hmm.

fast forward to linux again. Hes forcing a re-association everytime he wants to roam.
No wonder hes taking long!

Back to windows. Client is moved as smooth as butter.

Somethings not right on the linux side.

Drivers handling mesh roaming maybe? welp
 
Realize how this is 0.005% of reality -- connections are meant to be made for use and not just to see Can I and with how little loss. YES your variations between platforms are interesting, in the real world of usage patterns, it's a nit.
 
Back