[ https://issues.apache.org/jira/browse/SPARK-4359?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Andrew Or updated SPARK-4359: ----------------------------- Description: In the parent pom, "avro.mapred.classifier" is set to "hadoop2" for Yarn but not otherwise set. As a result, when an application that uses "spark-hive_2.10" as a module is built with SBT, it will try to resolve a jar that is literally called {code} avro-mapred-${avro.mapred.classifier}.jar {code} This is because "avro.mapred.classifier" is not a variable according to SBT. was: In the parent pom, "avro.mapred.classifier" is set to "hadoop2" for Yarn but not set otherwise. As a result, when a application built with SBT tries to use "spark-hive_2.10" as a module, it will try to resolve a jar that is literally called {code} avro-mapred-${avro.mapred.classifier}.jar {code} This is because "avro.mapred.classifier" is not a variable according to SBT. > Empty classifier in "avro-mapred" is misinterpreted in SBT > ---------------------------------------------------------- > > Key: SPARK-4359 > URL: https://issues.apache.org/jira/browse/SPARK-4359 > Project: Spark > Issue Type: Bug > Components: Build > Affects Versions: 1.1.0, 1.2.0 > Reporter: Andrew Or > Assignee: Andrew Or > Priority: Critical > > In the parent pom, "avro.mapred.classifier" is set to "hadoop2" for Yarn but > not otherwise set. As a result, when an application that uses > "spark-hive_2.10" as a module is built with SBT, it will try to resolve a jar > that is literally called > {code} > avro-mapred-${avro.mapred.classifier}.jar > {code} > This is because "avro.mapred.classifier" is not a variable according to SBT. -- 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