[IMPALA-8956] Row count incorrect in summary while query running

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Epic Color: ghx-label-8
• Assignee: Unassigned
• Reporter: Peter Ebert
• Created: 19/Sep/19 01:29
• Updated: 14/May/20 20:19
• Resolved: 14/May/20 20:19

Description

For a query (below) the summary is incorrect on the row count for the nested loop join, however when the query is complete its corrected:

select g.name, g.start, g.aa, sum(case when p.pn < 125.0 then 1 else 0 end) / cast(max(lf_count) as decimal(10,6)) as lf, sum(case when p.pn >= 125.0 then 1 else 0 end) / cast(max(hf_count) as decimal(10,6)) as hff from g join p on (g.sampleid = p.sample_id) cross join (select count as lf_count from p where pn < 125.0) lff cross join (select count as hf_count from p where pn >= 125.0) hf group by g.name, g.start, g.aa order by name desc

 

compared to the finished query

 

 

Issue Links

• duplicates: IMPALA-8026 Actual row counts for nested loop join are way too high while the query is executing