Imagine we have an item (monitored via proxy) and a nodata(120) trigger for it. Then following happens:
2014-07-30 17:00:00 - maintenance period starts
2014-07-30 17:03:08 - item value comes, trigger is OK
2014-07-30 17:04:53 - another value comes, but proxy delays it, so it does no get to server (this is achieved by temporarily setting wrong server ip in proxy config).
2014-07-30 17:05:30 - nodata(120) happens and trigger switches to PROBLEM
2014-07-30 17:05:3x - proxy delivers the second value, trigger switches to OK

As a result, event table is db has the following:

1. eventid, objectid, from_unixtime(clock), value, acknowledged
   '8006', '14134', '2014-07-30 17:04:53', '0', '0' <<<< record with bigger eventid has smaller time
   '8005', '14134', '2014-07-30 17:05:30', '1', '0'

Then values keep coming, trigger stays in OK condition.
Then 2014-07-30 17:10:00 - maintenance period ends and server generates a new PROBLEM event, that is actually wrong, since trigger is in OK condition.