Anyone know what the return time is for an AP to jump back to the primary controller from the secondary controller once it comes back online? I have a backup controller over a WAN connection that I'm using to backup four different locations. WLC is runnign 3.2.78 code. APs are 1000 series.
Well, I specified the IP Address in the secondary controller field, which is supposed to be SNMP name, but it works. I found that the fallback was impeded by my WAN. The APs were trunked on the switchport and the two sites I had problems with were using Cisco 340 series wireless bridges which do not pass VLAN tagged traffic since they are limited to 1500 bytes. When I set the switchports to access ports for the APs, this fixed the issue of them not connecting to the secondary. Since doing that, the fail over is about 10 - 15 seconds across my WAN.
We're trying to figure this out as well; we have a WiSM, and try to get the APs to fail over from one side to the other. They do, but it takes them a good 30 seconds and obviously no traffic is passed during this time. Both controllers share a mobility group, and each mobility group peer can ping the other.
The APs are converted 1131AGs an 12xx series. We've used http://www.cisco.com/warp/public/102/wlc_failover.pdf to set this up, but there reregistration is hardly immediate, and the APs don't seem to ever switch back to the primary controller once it comes back up. Any suggestions?
I think AP fallback needs to be enabled for the AP to return to the original WLC. Seems like I've heard 45-60 seconds for AP fail over to mobility peer. I'm going to test this in the lab with 2 4404 and 2 1242 and I'll post the results. With a console cable connected to the AP to watch the AP status, I wonder if the AP just does a LWAPP Join, or if it reloads the OS and does a new LWAPP Discover when the primary WLC goes away?
Beringer (4.0) code for the WLC is slated to be released in June-July. There may be some failover enhancements in the new code. I agree the current failover times are not acceptable for Wireless HA.
That time seems similar to what we're seeing, and is rather disappointing compared to the "instant" failover Cisco had claimed. Based on the console output the 1131 APs do a new discovery and join, but don't reload.
From my experiences, it seems that the Aironet LWAPP Ap's are not quite as responsive as the Airespace (1000 series) APs. I've had converted APs that I've had to reset to get them to move back from their secondary to primary. (Maybe I'm too impatient, we got tired of staring and waiting) This wasn't the case with the Airespace APs. They would move pretty quickly. I would expect that the lwapp code will continue to be refined via new releases and hopefully this will make for a quicker convergence.