Distinct/Count (aggregate ?) don't take limit into account

[FGRibreau]

select distinct(key) from monitoring limit 3;

Should returns 1 to 3 lines but instead returns every distinct key from the monitoring collection, it should not.

Same with count:

select count(key) from monitoring
┌───────────────┬───────┐
│ time          │ count │
├───────────────┼───────┤
│ 1395780644966 │ 3652  │
└───────────────┴───────┘
select count(key) from monitoring limit 40
┌───────────────┬───────┐
│ time          │ count │
├───────────────┼───────┤
│ 1395780644966 │ 3652  │
└───────────────┴───────┘

Tested on InfluxDB 0.5.0

[pauldix]

Note to @jvshahid: We should be processing the limit both at the shard level and at the coordinator level.

[jvshahid]

@FGRibreau your first example is considered a bug. When it comes to the second example, I think count() should return the number of all points and not just 40. This story will fix the behavior of limit to be applied after the aggregation is done, that means, limit will limit the number of points returned to the client and not the number of points that are input to the aggregator

[FGRibreau]

When it comes to the second example, I think count() should return the number of all points and not just 40

Indeed, the bug is only true for distinct. For reference in (my)SQL:

SELECT COUNT (id) FROM account LIMIT 40
// is indeed the same as `SELECT COUNT (id) FROM account`

// however
SELECT DISTINCT (id) FROM account LIMIT 40
// returns 40 lines. (all id from account are different)