[IMPALA-1124] Issue with Group By on a timestamp field

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: Impala 1.4, Impala 1.4.1
• Fix Version/s: Impala 2.0
• Labels: field groupby timestamp
• Environment: CDH5.1.0 Debian 7.X Impala 1.4.0 Parcel
• Assignee: Alexander Behm
• Reporter: Nicolas PHUNG
• Created: 31/Jul/14 14:12
• Updated: 20/Dec/15 00:05
• Resolved: 10/Sep/14 07:05

Description

SELECT subject,
IF(mail_sent_date_ts BETWEEN "2012-04-01 00:00:00" AND "2012-04-30 00:00:00", sent, 0) AS sent2
FROM mail 
WHERE activity_ts BETWEEN "2012-04-01 00:00:00" AND "2012-04-30 00:00:00"
GROUP BY subject, mail_sent_date_ts, sent

This request works on Impala 1.2.3 but not on Impala 1.4.0. I've got the following errors :

Your query has the following error(s):

AnalysisException: select list expression not produced by aggregation output (missing from GROUP BY clause?): IF(mail_sent_date_ts BETWEEN "2012-04-01 00:00:00" AND "2012-04-30 00:00:00", sent, 0)

I don't know if it's related but the field `mail_sent_date_ts` is a `timestamp` in Impala Parquet format.