Created on 03-07-2016 06:27 AM - edited 09-16-2022 03:07 AM
Hello frndz,
Can you please help me to understand below error.
Actually when I am doing union all join with two tables and creating one table with the help of this join then I am getting below error on tez execution engine.
Failed with exception MetaException(message:Invalid partition key & values; keys [feed_date, ], values []) FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.MoveTask
But when I am running it on mr then it is running fine. I am using hdp2.3.
SQL:
hive> INSERT OVERWRITE TABLE clickstream_kpis.cs_android_event_lookup_tbl PARTITION (feed_date)
> SELECT a.*
> FROM
> (SELECT hitid_high,
> hitid_low,
> event_number,
> feed_date
> FROM clickstream_db.clickstream_android LATERAL VIEW explode(SPLIT (post_event_list,','))expld AS event_number
> WHERE feed_date BETWEEN '2016-01-01' AND '2016-01-31'
> AND post_event_list IS NOT NULL
> UNION ALL SELECT hitid_high,
> hitid_low,
> post_event_list AS event_number,
> feed_date
> FROM clickstream_db.clickstream_android
> WHERE feed_date BETWEEN '2016-01-01' AND '2016-01-31'
> AND post_event_list IS NULL)a;
Query ID = hdpbatch_20160304145048_5262fb30-6ed6-4a7c-ad5d-ca30a1bc57d6
Created 03-07-2016 06:49 AM
Created 03-07-2016 06:53 AM
Yes @Neeraj Sabharwal: I am using Hive 1.2.1 and it seems it has been fixed in Hive 1.2.1 and higher version as per jira.But I am getting this error.
Created 03-07-2016 07:16 AM
@Saurabh Kumar You may have to check the data...There is a chance that there is no value for feed_date
Created 03-07-2016 07:47 AM
@Neeraj Sabharwal : I have checked and it has data as it is running with mr execution engine.
Created 03-07-2016 07:56 AM
@Saurabh Kumar You may want to open a support ticket and update that jira with your findings.
Created 03-07-2016 10:10 AM
@Neeraj Sabharwal: Thanks Neeraj,
I have opened a case and working on it with support team. Let see what would be finding, I will update here if I will find solution.
Created 08-03-2016 05:05 AM
The issue can be resolved in lower versions also by using the distribute by keyword in the query