1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

WDS: Single Failure Requires Full Network Restart

Discussion in 'Tomato Firmware' started by cray54, Dec 7, 2013.

  1. cray54

    cray54 Reformed Router Member

    When using WDS with two WRT54G(S) routers, if one router loses power, both routers need to be restarted to re-establish the connection. Also, a WET54G connected to the network needs to be restarted while WRT54G(S) routers are being restarted to allow it to connect back to the network.

    Is there a configuration that allows the WDS system to re-connect after a failure? The routers are in different buildings and it is challenging to restart them all at the same time.

    I am using a WRT54G v1 and WRT54GS v1 both running:
    Tomato Firmware v1.28.7635 Toastman-IPT-ND ND VLAN-VPN

    Thanks, Chris
  2. Marcel Tunks

    Marcel Tunks Networkin' Nut Member

    You can try changing WDS from "link with" to "automatic" on the basic settings page.
  3. mvsgeek

    mvsgeek Addicted to LI Member

    Couple of questions...

    1. Which encryption scheme are you using? If WPA2, try WPA instead.
    2. Why are you mixing WDS and WET? Does the WET router connect to your 'main' (WAN) router or the secondary?
    3. How far apart are the 2 routers?

    In my limited experience, when dealing with WDS connections, timing is everything. And the more connections, the less reliable the network. Over the years, several people with similar configurations have posted 'keepalive' type scripts on this forum (raises hand).
    Marcel Tunks likes this.
  4. cray54

    cray54 Reformed Router Member

    Murcel -
    I will experiment with the "automatic" setting. It would likely give me routing problems in the future since I would like to add WDS points, but it could solve the current problem.

    mvsgeek -
    1. I am using WPA2. I will experiment with WPA (to help solve if encryption is the problem) but will not use it as a solution.

    2. The WET54G is just a client on a wireless network (not another router). I use it to allow a wired device to connect wirelessly to my network. It happens to connect to the 'main' router but has worked when connecting to the other WDS node.

    3. The WDS nodes are ~480 ft apart with clear line of sight. SNR is around 19 db.

    This setup is very very stable when linked. I can run it for weeks without problems and maintain <2 ms pings across the WDS link.

    The only problem at this time is when one node loses power while the other maintains it. After restoring power, the "Device List" will show the "wds0.1" link but no communication will be possible. This remains true until both units are restarted at the same time.
  5. Goggy

    Goggy Network Guru Member

    i'm using WDS on 3 WRT54G's for a few years and over all its working flawlessy. Like you said correctly, the only problem are power - losses. As a workaround for this problem i use a "keepalive-script" which i put in administration/script/init:
    cat <<END > /tmp/keepalive
    ping -c5 >/dev/null || reboot
    chmod 755 /tmp/keepalive
    cru a keepalive "*/2 * * * * /tmp/keepalive >/dev/null 2>&1"
    Every 2 minutes the other wds-node is pinged 5 times, if there is no response the WRT54G reboots. This script - with adjusted ip's - runs on every wds-node. Personally i would not use "wds automatic", i would "link with" ...
    cray54 likes this.
  6. mmosoll

    mmosoll Serious Server Member

    Thank you, same problem with E4200. Using script with 5-8 minutes interval works fine, 2 minutes were too short
  7. cray54

    cray54 Reformed Router Member

    Is there a service or process that can be restarted to achieve the same goal? I'd like my wired hardware to remain connected during the repair if at all possible.
  8. cray54

    cray54 Reformed Router Member

    This has been working very well with my two WRT54G(S)s. I even added a third to the mix running DD-WRT and turned on the WDS Watchdog for 2 minutes as well. Any node that goes down for too long now restarts all of the nodes and the network continues to function!

    I would still like to see a solution that does not take down the wired clients, but this is a huge step for my system.
  9. mvsgeek

    mvsgeek Addicted to LI Member

    You could try replacing the 'reboot' command in your keepalive script with one or more 'wl radio' commands...

    When used without arguments, toggles radio on/off.

    up - reinitialize adapter and mark as up
    down - reset adapter and mark as down
    out - mark adapter down but do not reset hardware
    restart -

    wl radio
    wl radio up
    mmosoll likes this.

Share This Page