How to select backup WDS AP

[yueric]

I'm curious about how to select a WDS AP from multiple APs that has the same priority number?

[#2]

You should not have multiple WDS APs with the same priority. There should be one WDS Master (with highest priority) and a maximum of two other APs with lower priority for backup (again, the backups should not have the same priority as the primary and the backups should not have the same priority as each other).

[yueric]

Thanks for your reply,but I'm curious what will it happen when multiple APs with the same priority.

[dewman03]

If multiple AP's have the same WDS proirity

Then there will be a random WDS master chosen.

[derekliedl]

Why shouldn't there be two with lower priority and no more? I currently have a site with 60+ AP's. There is of course a primary WDS, but each AP can function as a backup based on priority. The primary has a priority of 250, first backup is 248, second backup is 246, all the way down the chain. I've never had a problem with thi setup. So other than the inconvenience of keeping track of all the info, is there a downside to having multiple backup WDS ap's?

[jmagnusson]

Have to tried this..? if you have 30 AP's in a WDS domain - How do you have 1 master and a backups and have the othe 27 not participate in the WDS master election process ? When I shut down my master - I have a 30 sec to 1 min outage, until my configured backup takes over. 1 master - Priority 250 . 1 Backup - Priority 200 - 28 Backups' Priority 100.

Cant figure out how to make the 28 backups - not be candidates.

[derekliedl]

Not sure I follow, but here is what I have:

One master with a priority of 250. When/if that AP dies, the AP with a priority of 248 takes over. When that AP dies, the AP with a priority of 246 takes over, on down the chain. Now, that being said I've only ever tested what happens when 250 goes down, never further down the chain. But the jump from 250 as a master to 248 becoming the master is seamless, at least using IOS 15JA. We had some issues with the long outage up to a minute using previous firmwares, but on the newest it's been seamless.

[jmagnusson]

So if you shut off you master while you are authenicated.. and you are logged into your 248 candidate.. How long before that AP is now showing as the active wds ?

[derekliedl]

I can't recall exactly. Right now I don't have enough AP's and devices in my lab to quick test it, and I can't do it in a production area because, well, because it's a production area.

But when I've done it in the past, it's happened fast enough that clients did not drop. For example: Recently we had a warehouse go live on Cisco wifi. This was about 50 AP's throughout a massive building, about 500 yards end to end, with multiple floors. The clients are handhelds and forklift-mounted terminals, roughly 60 in total.The day we started testing this we wound up having a routing issue, but until that was discovered I thought perhaps it was a WDS problem. I shut down the primary WDS so the secondary could take over, and all our hardware guys that were tooling around on lifts, testing coverage, etc. never even noticed. I did this several times trying to pinpoint the problem. I shut down the highest priority WDS to let the second highest take over, then brought the highest priority back up to take control again, and we didn't miss a beat. I even manually changed the priority on the primary WDS from 250 down to 5 at one point, and even doing that the transition to the second highest priority seemed seamless.

Now bear in mind that this was using 15JA firmware. I had problems in the past with previous firmware versions causing the primary WDS to randomly reboot no matter which AP it was on, and this tended to cause like a minute drop where everything is messed up. Since we upgraded to 15JA that dissapeared. In that case, the lag wasn't the leap from primary to backup WDS, it was the firmware bug.

[jmagnusson]

Thanks for the info but that is now what I saw. We are running EAP-FAST on CIsco B cards with 15JA IOS on the APs. When we shut down the master and we walked where we should roam we did not reauthenicate for 30 secs or so. You do know the olny time you would see a problem if you roamed at the time time the WDS master election process was taking over. If you arent roaming you will not see it...

I wish we could shut off 28 of our 30 APs in the one WDS domain to see if the election process was sped up.

[derekliedl]

Hmmm... well, we are using LEAP and not EAP-FAST, and on top of that we are using local radius authentication internally in the AP instead of at a Radius server, so that could definitely account for our different results.