Hello

I hope someone can help throw some light on this subject. We have a large network with somewhat unusual topology.
We have approx. 40 Routerboards linked in a big T topology, inside a hydro power station project. All routerboards are bridged AP+bridge, and most link via WDS
The network diameter is 18 hops and we are starting to experience some problems. When adding new AP's at one of the extremes (as the tunnel drifting is progressing), APs in the other extreme becomes inaccessible. The WDS link will still be up, but its not forwarding data. At certain intervals all data traffic freezes in the entire network.

I suspect the problem is related to the increasing network diameter and our using STP default parameters.
Does anyone have any experience with adjusting the STP parameters? I've found a pretty good Cisco document (Understanding and tuning spanning tree protocol timers), but the site is not ideal for experiments (narrow tunnels, big heavy rock carrying trucks in steep slippery slopes )

Introducing routing is less attractive, as I have little experience, and I will have to reconfigure a rather large number of WLAN phones (with no web interface), and other equipment.
Any input is welcome

Regards,
MortenK

Maybe switch to RSTP?
Also, check the wikipedia page for STP, it has some useful info.

If you have no loops/rings in your network, maybe turn off STP all together?

Hi, thanks for the input.

We are using rSTP.

I will try turning it off as the first step. Why complicate things! Thanks.

Regards,
Morten

Hi, I too have had this problem. On the 5th hop of a long wds run I have this exact problem. The temp solution was to use PSudobridge mode which works OK with it's limitations, but also I have same prob. Can't make work with wds? any new info? I tried the different rstp protocols with no difference?

Do you used single or multiple radio on each AP. what's the latency between one end to the other. whats the traffic like.

used multiple radio , one for uplink, one for downlink, one to serve your user. the can still be in bridge mode.

The steup i'm using two radios on each relay one wds station mode and one in AP mode to realy the network. Speed and latency is good. Usually 3 ms accross all the hops which actually goes possibly 60 miles. Just this little wds weird thing?

Update:
Disabled rSTP on all AP's on each side of T-junction. All AP's from the junction and south, towards the server have rSTP enabled. Now been stable for 12 hours, we are about to leave the site.
Our setup is 2 radios pr Mikrotik device, both in AP+ bridge mode. Each radio serves as either up- or downlink, + serves our WLAN phones. Some devices have 3 radios for additional WLAN coverage.
We have now 20 hops on the longest leg, from server to last AP on the western branch. Physically its about 3-4km, most of it inside the mountain.
Latency is good enough to not be disruptive for WLAN telephony, ping average @60ms

Regards,
Morten

have a similar problem when I begin to use RSTP + WDS, every 30s-60s the entire network freeze.

Look like the problem is the dynamic wds cost, you have a wds link that change the cost every a few seconds, this force a recalculation of the RSTP root and freeze the entire network.

check your WDS Cost Range, change it from the default 50-150 to 100 in every link.

using winbox, change the interface wds settings (in advanced mode)

leave the rest of the RSTP parameter at the default values, but make sure that you set the correct priority for the bridge that you want to be the root

Only changing this value you can stop this freeeze of your network?


I have the same problem in a network with around 40 RBs in the city with dual radio each... 5Ghz(wds-mesh, rstp, nstreme) to inter-nodes links and the other radio like a simple AP 2.4Ghz for more than 1200 customers...

I have been configured the priority in the network, all centralized in a Hotspot Server with priority 1(root bridge)

Thanks and regards!