Steps to reproduce:

1. Create an item, using discard unchanged preprocessing, where the item value does not change regularly

1. 1. eg: vmware.version[]
2. Wait for item to populate
3. Cause item to fail (restart server, shutdown service port/etc)
4. Allow item to recover
5. Wait for polling interval
6. Item will stay in failed status
7. Remove discard unchanged preprocessing rule
8. Item will recover

Expected:
An item would recover after being successfully polled.  I assume that success status is only updated if data is entered into the DB.