Imagine we have an item (monitored via proxy) and a nodata(120) trigger for it. Then following happens:
2014-07-30 17:03:08 - first item value comes, trigger switches to OK
2014-07-30 17:04:53 - second 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:

# eventid, objectid, from_unixtime(clock), value, acknowledged
'8006', '14134', '2014-07-30 17:04:53', '0', '0'
'8005', '14134', '2014-07-30 17:05:30', '1', '0'

Now see the attached screenshot. The triggers page on the left says (correctly):
PROBLEM 2014-07-30 17:05:30 Duration 13s
The events page on the right says:
2014-07-30 17:05:30 no data from foo PROBLEM Duration 37s
The 37s here is incorrect - it's a duplicate from below line
2014-07-30 17:04:53 no data from foo OK Duration 37s