Github user vanzin commented on a diff in the pull request: https://github.com/apache/spark/pull/22624#discussion_r229032985 --- Diff: core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedSchedulerBackend.scala --- @@ -230,7 +242,7 @@ class CoarseGrainedSchedulerBackend(scheduler: TaskSchedulerImpl, val rpcEnv: Rp val reply = SparkAppConfig( sparkProperties, SparkEnv.get.securityManager.getIOEncryptionKey(), - fetchHadoopDelegationTokens()) + Option(delegationTokens.get())) --- End diff -- Although it's not obvious I think this is fine. The token manager is started before any rm-specific code runs and requests executors (in the `CoarseGrainedSchedulerBackend.start()` method; rm-specific schedulers call that before they do anything). So the message to set these tokens is the first message queued on the driver, and this reference will already be set when the first executor asks for this info.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org