I've seen that before, your best bet and fastest way to fix this is to separate the clusters, upgrade them separately and join them back.
Think of the upgrade being "stuck". One of your mixed model APs is not fully upgraded.
If you know the IP address of each IAP, you can ssh into each of them and type the following:
show version #which version the AP is on
show image version #which version each partition has (see if your AP has been pushed the correct firmware)
upgrade-image (then hit tab), you will see:
* upgrade-image
* upgrade-image2
* upgrade-image2-no-reboot
* upgrade-image2-no-switch-partition-reboot
Try upgrading to the correct image, if there using the command above, if not, do the first step.