I noticed that after an application restart on the 25th of Jan ( maybe during trending) the data that should have been aggregated that day is missing. I was kind of expecting zabbix to re-run the missing part instead of just skipping over it.

Is this a bug or a known limitation of the system?

Here is an example of an item that only getting new data once a day and has no data in trends for the 25th:

SELECT FROM_UNIXTIME(t.clock, '%Y %D %M %h:%i:%s %x') clock
, t.value_min
, t.value_avg
, t.value_max
FROM trends t
, items i
WHERE t.itemid=i.itemid
AND i.key_='some.custom.item';

2011 19th January 12:00:00 2011 54.0 54.0 54.0
2011 20th January 12:00:00 2011 40.0 40.0 40.0
2011 21st January 12:00:00 2011 58.0 58.0 58.0
2011 22nd January 12:00:00 2011 57.0 57.0 57.0
2011 23rd January 12:00:00 2011 100.0 100.0 100.0
2011 24th January 12:00:00 2011 100.0 100.0 100.0
2011 26th January 12:00:00 2011 59.0 59.0 59.0

====================================================================
And to prove the data is in the raw data table:

SELECT FROM_UNIXTIME(h.clock, '%Y %D %M %h:%i:%s %x') clock
, h.value
FROM history h
, items i
WHERE h.itemid=i.itemid
AND i.itemid=62239;

2011 19th January 12:05:24 2011 54.0
2011 20th January 12:05:23 2011 40.0
2011 21st January 12:05:24 2011 58.0
2011 22nd January 12:05:23 2011 57.0
2011 23rd January 12:05:23 2011 100.0
2011 24th January 12:05:26 2011 100.0
2011 25th January 12:05:27 2011 70.0 <-----------------------------------
2011 26th January 12:05:24 2011 59.0
2011 27th January 12:05:25 2011 61.0