[ https://issues.apache.org/jira/browse/SPARK-5861?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sean Owen updated SPARK-5861: ----------------------------- Fix Version/s: (was: 1.2.2) (was: 1.3.0) Ah, you're saying that you are in yarn-client mode but the Application Master is still using the large amount of memory defined in spark.driver.memory, instead of the small default of 512m it should use. I see that the lines of code you cite only execute in yarn-cluster mode though, or should only execute. That in turn is keyed on whether {{--class}} is set by {{Client.scala}} which again should only happen in yarn-cluster mode. How are you running this? are you bypassing this code? CC [~sandyr] and [~vanzin] for an opinion. > [yarn-client mode] Application master should not use memory = > spark.driver.memory > --------------------------------------------------------------------------------- > > Key: SPARK-5861 > URL: https://issues.apache.org/jira/browse/SPARK-5861 > Project: Spark > Issue Type: Bug > Components: YARN > Affects Versions: 1.2.1 > Reporter: Shekhar Bansal > > I am using > {code}spark.driver.memory=6g{code} > which creates application master of 7g > (yarn.scheduler.minimum-allocation-mb=1024) > Application manager don't need 7g in yarn-client mode. -- 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