Michael Brown created IMPALA-6978:
-------------------------------------
Summary: TPCH Q11 in testdata is not SF-aware
Key: IMPALA-6978
URL: https://issues.apache.org/jira/browse/IMPALA-6978
Project: IMPALA
Issue Type: Bug
Components: Infrastructure
Affects Versions: Impala 2.8.0
Reporter: Michael Brown
[http://www.tpc.org/tpc_documents_current_versions/pdf/tpc-h_v2.17.3.pdf] p47
mentions Q11 is:
{noformat}
select
*
from (
select
ps_partkey,
sum(ps_supplycost * ps_availqty) as value
from
partsupp,
supplier,
nation
where
ps_suppkey = s_suppkey
and s_nationkey = n_nationkey
and n_name = 'GERMANY'
group by
ps_partkey
) as inner_query
where
value > (
select
sum(ps_supplycost * ps_availqty) * 0.0001 / SF
from
partsupp,
supplier,
nation
where
ps_suppkey = s_suppkey
and s_nationkey = n_nationkey
and n_name = 'GERMANY'
)
order by
value desc
{noformat}
{{testdata/workloads/tpch/queries/tpch-q11.test}} has this hardcoded to imply
SF==1, which is fine for our minicluster, but not fine for tests that use code
upstream and run in downstream environments on large SFs, like the stress test.
On TPCH 10,000 for example this query produces no results. If you apply the
correct formula with SF, though, this does produce results.
Possible solutions: dynamically write the query based on SF, or keep a few
queries at well-known scale factors.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
