Can you paste the queries here, along with what di...
# troubleshooting
a
Can you paste the queries here, along with what discrepancy are you seeing?
a
1. Query which is trimmed select dim1,sum(metric1) metric1,sum(metric2) metric2 FROM table1 WHERE statsdatehour_epoch between 1651449600000 AND 1651881599000 AND event_status = 1 AND partner_id IN (1012,1057,1058,1059,1090,1157,1188) GROUP BY dim1 ORDER BY metric1 DESC LIMIT 500 option(minServerGroupTrimSize=-1)
2. Actual value for a group which is trimmed. select dim1 ,sum(metric1) metric1,sum(metric1) metric1 FROM table1 WHERE statsdatehour_epoch between 1651449600000 AND 1651881599000 AND event_status = 1 AND partner_id IN (1012,1057,1058,1059,1090,1157,1188) and lower(dim1) = '/3865/peo.mdp.amp/tier2/article/country' GROUP BY dim1 ORDER BY metric1 DESC LIMIT 500 option(timeoutMs=60000)
image.png
For this group, '/3865/peo.mdp.amp/tier2/article/country' values are different.
r
do you still get discrepancies if you add an order by clause?
s
Yes, even with order by discrepancies are observed. We get clean data only when DIM value is explicitly in filter clause(query 2 mentioned by anish)
r
what about order by dim1?
s
yes, same issue on order by dim1 too we can see.