yes, try organize all your queries with a "group by" On Fri, Mar 6, 2015 at 10:30 AM, dong wang <[email protected]> wrote:
> if ""select my_date from test group by mydate order by my_date desc limit > 10", the result is correct > > 2015-03-06 10:25 GMT+08:00 hongbin ma <[email protected]>: > > > can you rewrite your query "select my_date from test order by my_date > desc > > limit 10" to "select my_date from test group by mydate order by my_date > > desc limit 10" > > > > for the reason please refer to what i replied to Santoshakhilesh > > > > On Fri, Mar 6, 2015 at 10:22 AM, dong wang <[email protected]> > wrote: > > > > > Hi hongbin, just as I mentioned in the previous email, I built the cube > > > with just one big fact table, so all the queries are related to the > fact > > > table, and just now this morning, still get incorrect result as listed > > > above with REST API even though I set "acceptPartial = false" for the > > query > > > parameters > > > > > > 2015-03-06 10:05 GMT+08:00 hongbin ma <[email protected]>: > > > > > > > hi, dong, > > > > > > > > If a query is not related to fact table, Kylin will not leverage > cubes > > to > > > > answer the query. Instead, it will use the lookup table's snapshot. > > > > > > > > In other words, query execution logic changes when you're querying > > lookup > > > > table only. can you please try using "facttable join lookup" pattern > in > > > > your query? and meantime I'll check if there's any bug in the > snapshot > > > > approach. > > > > > > > > thanks > > > > hongbin > > > > > > > > > >
