I have created a partitioned index as below. the actual index is different, I am giving a similar example with travel-sample
CREATE INDEX `idx_grp_add` ON `travel-sample`((`geo`.`alt`),(`geo`.`lat`),(`geo`.`lon`),`id`) PARTITION BY HASH((`geo`.`alt`),(`geo`.`lat`),(`geo`.`lon`),`id`)
Then, I ran the below query
SELECT geo.alt, geo.lat, SUM(geo.lon), AVG(id), COUNT(DISTINCT geo.lon) FROM `travel-sample` WHERE geo.alt BETWEEN 10 AND 30 GROUP BY geo.alt, geo.lat HAVING SUM(geo.lon) > 1000;
I was expecting that the GROUP BY operation will be pushed to the indexer. But I see it has not. if I do not have a partitioned index, the GROUP BY operation does get pushed to indexer. Is it because the index is partitioned, the data comes from different partitions and the query node has to do the group by because Indexer does not have all the portions of the index data spread across the partitions.