Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
Description
git.commit.id.abbrev=6f54223
I have a hive table on top of a parquet file. When we have a type mis-macth for any column in the hive ddl, the queries from drill fail. However there seems to be an issue with error propogation
In embedded mode the error seems to be propogated as expected
0: jdbc:drill:zk=local> select * from customer;
Error: SYSTEM ERROR: java.lang.ClassCastException: org.apache.hadoop.hive.ql.io.parquet.writable.BinaryWritable cannot be cast to org.apache.hadoop.io.IntWritable
Fragment 0:0
[Error Id: 459c71ea-c66e-459b-9d66-8408b4bf0954 on qa-node190.qa.lab:31010] (state=,code=0)
In a multi-node setup, there is not message on sqlline
0: jdbc:drill:schema=dfs_eea> select count(*) from customer; Error: SYSTEM ERROR: Fragment 0:0 [Error Id: e5ac4048-73fa-441a-b6a7-fb1e25838d04 on qa-node191.qa.lab:31010] (state=,code=0)
Log files in a multi-node setup also do not contain the message. I attached the logs and embedded and multi-node setup along with the data. Below is hive ddl (wrong type for phone)
create external table if not exists tpch01_parquet_nodate.customer ( c_custkey int, c_name string, c_address string, c_nationkey int, c_phone int, c_acctbal double, c_mktsegment string, c_comment string ) STORED AS PARQUET LOCATION '/tpch/customer/customer.parquet';
Let me know if you need anything
Attachments
Attachments
Issue Links
- is duplicated by
-
DRILL-3208 Hive : Tpch (SF 0.01) query 10 fails with a system error when the data is backed by hive tables
- Resolved