Shaofeng SHI created KYLIN-2543:
-----------------------------------
Summary: Still build dictionary for TopN group by column even
using non-dict encoding
Key: KYLIN-2543
URL: https://issues.apache.org/jira/browse/KYLIN-2543
Project: Kylin
Issue Type: Bug
Components: Job Engine
Affects Versions: v2.0.0
Reporter: Shaofeng SHI
Assignee: Shaofeng SHI
Fix For: v2.0.0
Using latest 2.0 beta, create a cube with TopN, but specifying using non-dict
encoding, the JSON is like:
{code}
{
"name": "TOP_CUSTOMER",
"function": {
"expression": "TOP_N",
"parameter": {
"type": "column",
"value": "P_LINEORDER.V_REVENUE",
"next_parameter": {
"type": "column",
"value": "P_LINEORDER.LO_CUSTKEY"
}
},
"returntype": "topn(100,4)",
"configuration": {
"topn.encoding.P_LINEORDER.LO_CUSTKEY": "integer:4",
"topn.encoding_version.P_LINEORDER.LO_CUSTKEY": "1"
}
}
{code}
But, in the third step "Extract Fact Table Distinct Columns", there is still a
reducer running for "P_LINEORDER.LO_CUSTKEY"; as this column is a UHC, it takes
a long time and finally got failed.
Expected behavior: no reducer for this column. 1.6.0 doesn't have this issue;
this is a new issue in 2.0 during the upgrade to snowflake.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
