[ https://issues.apache.org/jira/browse/SPARK-32700?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17184175#comment-17184175 ]
Sean R. Owen commented on SPARK-32700: -------------------------------------- I don't think you're guaranteed to get exactly 50% here. With a small data set, due to random sampling, it could be even 3 with not-small probability. > select from table TABLESAMPLE gives wrong resultset. > ---------------------------------------------------- > > Key: SPARK-32700 > URL: https://issues.apache.org/jira/browse/SPARK-32700 > Project: Spark > Issue Type: Bug > Components: SQL > Affects Versions: 3.0.0 > Environment: Spark 3.0.0 > Reporter: Chetan Bhat > Priority: Minor > > create table test(id int,name string) stored as parquet; > insert into test values > (5,'Alex'),(8,'Lucy'),(2,'Mary'),(4,'Fred'),(1,'Lisa'),(9,'Eric'),(10,'Adam'),(6,'Mark'),(7,'Lily'),(3,'Evan'); > SELECT * FROM test TABLESAMPLE (50 PERCENT); --> output is giving only 3 rows. > spark-sql> SELECT * FROM test TABLESAMPLE (50 PERCENT); > 5 Alex > 10 Adam > 4 Fred > > Expected as per the link is 5 rows > -->[https://spark.apache.org/docs/latest/sql-ref-syntax-qry-select-sampling.html] > > Also the bucket parameter for select from table TABLESAMPLE gives wrong > resultset. > spark-sql> SELECT * FROM test TABLESAMPLE (BUCKET 4 OUT OF 10); > 5 Alex > 8 Lucy > 9 Eric > 1 Lisa > 3 Evan > Expected is 4 records. > -- 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