I have a iscsi datastore that reports as over allocated. The alarm set to 100% for warning. But the actual size is not even near to be over allocated.
There is no option to reset to green but only to acknowledge.
Tried refresh and restrting vcenter service.
A couple of things to check:
Is it possible that there are leftover files on the datastore for VMs that don't reside there anymore, such as old .vmdks taking up space? You can open up the Browse Datastore browser and compare what you see listed against what is actually running and living on this datastore.
Maybe there is another overallocation alarm being applied from a different level in the vCenter tree? I've seen this in the past, especially after upgrades to vCenter Server. Having a default overallocation under 100% would be odd though...
Sometimes an alarm will get stuck, I especially have found this issue with storage alarms (such as datastore usage). I have found that this trick works:
See if that clears it up.