[ https://issues.apache.org/jira/browse/SPARK-30559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17030083#comment-17030083 ]
Dongjoon Hyun commented on SPARK-30559: --------------------------------------- Thank you for reporting with the detailed information. Could you check the other Spark versions (e.g. 2.4.3 or 2.3.4, or 3.0.0-preview2), too? BTW, there are something you should consider. 1. Apache Hive is case insensitive while Apache Parquet is case sensitive. 2. Apache Hive considers all columns nullable while Parquet doesn't. Especially due to (1), I usually don't recommend to mix the upper and lower cases. > Spark 2.4.4 - spark.sql.hive.caseSensitiveInferenceMode does not work with > Hive > ------------------------------------------------------------------------------- > > Key: SPARK-30559 > URL: https://issues.apache.org/jira/browse/SPARK-30559 > Project: Spark > Issue Type: Bug > Components: SQL > Affects Versions: 2.4.4 > Environment: EMR 28.1 with Spark 2.4.4, Hadoop 2.8.5 and Hive 2.3.6 > Reporter: Ori Popowski > Priority: Major > > In Spark SQL, spark.sql.hive.caseSensitiveInferenceMode INFER_ONLY and > INFER_AND_SAVE do not work as intended. They were supposed to infer a > case-sensitive schema from the underlying files, but they do not work. > # INFER_ONLY never works: it will always user lowercase column names from > Hive metastore schema > # INFER_AND_SAVE only works the second time {{spark.sql("SELECT …")}} is > called (the first time it writes the schema to TBLPROPERTIES in the metastore > and subsequent calls read that schema, so they do work) > h3. Expected behavior (according to SPARK-19611) > INFER_ONLY - infer the schema from the underlying files > INFER_AND_SAVE - infer the schema from the underlying files, save it to the > metastore, and read it from the metastore on any subsequent calls > h2. Reproduce > h3. Prepare the data > h4. 1) Create a Parquet file > {code:scala} > scala> List(("a", 1), ("b", 2)).toDF("theString", > "theNumber").write.parquet("hdfs:///t"){code} > > h4. 2) Inspect the Parquet files > {code:sh} > $ hadoop jar parquet-tools-1.11.0.jar cat -j > hdfs:///t/part-00000-….snappy.parquet > {"theString":"a","theNumber":1} > $ hadoop jar parquet-tools-1.11.0.jar cat -j > hdfs:///t/part-00001-….snappy.parquet > {"theString":"b","theNumber":2}{code} > We see that they are saved with camelCase column names. > h4. 3) Create a Hive table > {code:sql} > hive> CREATE EXTERNAL TABLE t(theString string, theNumber int) > > ROW FORMAT SERDE > 'org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe' > > STORED AS INPUTFORMAT > 'org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat' > > OUTPUTFORMAT > 'org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat' > > LOCATION 'hdfs:///t';{code} > > h3. Reproduce INFER_ONLY bug > h4. 3) Read the table in Spark using INFER_ONLY > {code:sh} > $ spark-shell --master local[*] --conf > spark.sql.hive.caseSensitiveInferenceMode=INFER_ONLY{code} > {code:scala} > scala> spark.sql("SELECT * FROM default.t").columns.foreach(println) > thestring > thenumber > {code} > h4. Conclusion > When INFER_ONLY is set, column names are lowercase always. > h3. Reproduce INFER_AND_SAVE bug > h4. 1) Run the for first time > {code:sh} > $ spark-shell --master local[*] --conf > spark.sql.hive.caseSensitiveInferenceMode=INFER_AND_SAVE{code} > {code:scala} > scala> spark.sql("SELECT * FROM default.t").columns.foreach(println) > thestring > thenumber{code} > We see that column names are lowercase > h4. 2) Run for the second time > {code:scala} > scala> spark.sql("SELECT * FROM default.t").columns.foreach(println) > theString > theNumber{code} > We see that the column names are camelCase > h4. Conclusion > When INFER_AND_SAVE is set, column names are lowercase on first call and > camelCase on subsquent calls. > > -- This message was sent by Atlassian Jira (v8.3.4#803005) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org