[DRILL-2167] Order by on a repeated index from the output of a flatten on large no of records results in incorrect results

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Duplicate
• Fix Version/s: 1.1.0
• Component/s: Execution - Relational Operators
• Assignee: Jason Altekruse
• Reporter: Rahul Kumar Challapalli
• Created: 04/Feb/15 22:00
• Updated: 01/Jul/15 01:01
• Resolved: 26/Jun/15 17:55

Description

git.commit.id.abbrev=3e33880

The below query results in 200006 records. Based on the data set we should only receive 200000 records.

select s.uid from (select d.uid, flatten(d.map.rm) rms from `data.json` d) s order by s.rms.rptd[1].d;

When I removed the order by part, drill correctly reported 200000 records.

select s.uid from (select d.uid, flatten(d.map.rm) rms from `data.json` d) s;

I attached the data set with 2 records. I copied over the data set 50000 times and ran the queries on top of it. Let me know if you have any other questions

Issue Links

• duplicates: DRILL-2838 Applying flatten after joining 2 sub-queries returns empty maps
• is related to: DRILL-2232 Flatten functionality not well defined when we use flatten in an order by without projecting it
• is related to: DRILL-2238 Unsupported join on OR yields rough CannotPlanException (vs. cleaner unsupported-feature message)
• relates to: DRILL-2838 Applying flatten after joining 2 sub-queries returns empty maps
• relates to: DRILL-2083 order by on large dataset returns wrong results