turns out there is a bug in 6.4.2.9 + 6.4.2.10 + 6.3.1.17 that may have also caused this (affects ap 1xx models, it is a slow network buffer leak on the APs caused by client match. Eventually the AP becomes non responsive - it may manifest as all APs down)
Is there any possibility the APs were up for some time and then became down ? At any rate, please dont re-upgrade to 6.4.2.x until 6.4.2.11 is out (soon)
regards
-jeff