Github user mgummelt commented on a diff in the pull request: https://github.com/apache/spark/pull/17665#discussion_r112272077 --- Diff: core/src/main/scala/org/apache/spark/deploy/SparkSubmit.scala --- @@ -564,12 +566,22 @@ object SparkSubmit extends CommandLineUtils { // properties and then loaded by SparkConf sysProps.put("spark.yarn.keytab", args.keytab) sysProps.put("spark.yarn.principal", args.principal) - - UserGroupInformation.loginUserFromKeytab(args.principal, args.keytab) } } } + + // [SPARK-20328]. HadoopRDD calls into a Hadoop library that fetches delegation tokens with + // renewer set to the YARN ResourceManager. Since YARN isn't configured in Mesos mode, we + // must trick it into thinking we're YARN. + if (clusterManager == MESOS && UserGroupInformation.isSecurityEnabled) { + val shortUserName = UserGroupInformation.getCurrentUser.getShortUserName + val key = s"spark.hadoop.${YarnConfiguration.RM_PRINCIPAL}" --- End diff -- Is it? It looks like `getCurrentUser` returns the subject from the current `AccessControllerContext`, which is set by the `doAs` when `--proxy-user` is set. Regardless, the renewer specified here actually has no effect, since we aren't renewing yet. Once I add renewal, I'll need to revisit this to make sure it's consistent with the renewal we're going to do in `MesosSecurity`. I'm only setting this now to avoid an Exception that gets thrown in the hadoop library if the master principal is not configured. See the JIRA in the above code comment for details.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org