[
https://issues.apache.org/jira/browse/SPARK-15705?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15384464#comment-15384464
]
Michael Allman commented on SPARK-15705:
----------------------------------------
Yeah, I'm not too keen on that code path either. Inferring the schema from
reading every file in a partitioned table is a relatively heavyweight and slow
operation. We're not leveraging the fact that we have a metastore schema. This
is a performance/efficiency issue I've been working on in our own codebase, and
I believe we'll have something to contribute in the near future. However, that
will definitely not make it into 2.0. As for your problem specifically, I can't
really suggest a quick fix because I have no experience with the ORC file
format.
> Spark won't read ORC schema from metastore for partitioned tables
> -----------------------------------------------------------------
>
> Key: SPARK-15705
> URL: https://issues.apache.org/jira/browse/SPARK-15705
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 2.0.0
> Environment: HDP 2.3.4 (Hive 1.2.1, Hadoop 2.7.1)
> Reporter: Nic Eggert
> Priority: Critical
>
> Spark does not seem to read the schema from the Hive metastore for
> partitioned tables stored as ORC files. It appears to read the schema from
> the files themselves, which, if they were created with Hive, does not match
> the metastore schema (at least not before before Hive 2.0, see HIVE-4243). To
> reproduce:
> In Hive:
> {code}
> hive> create table default.test (id BIGINT, name STRING) partitioned by
> (state STRING) stored as orc;
> hive> insert into table default.test partition (state="CA") values (1,
> "mike"), (2, "steve"), (3, "bill");
> {code}
> In Spark
> {code}
> scala> spark.table("default.test").printSchema
> {code}
> Expected result: Spark should preserve the column names that were defined in
> Hive.
> Actual Result:
> {code}
> root
> |-- _col0: long (nullable = true)
> |-- _col1: string (nullable = true)
> |-- state: string (nullable = true)
> {code}
> Possibly related to SPARK-14959?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
