Github user mccheah commented on a diff in the pull request:
https://github.com/apache/spark/pull/19954#discussion_r157913013
--- Diff:
resource-managers/kubernetes/core/src/main/scala/org/apache/spark/scheduler/cluster/k8s/ExecutorPodFactory.scala
---
@@ -209,9 +213,33 @@ private[spark] class ExecutorPodFactoryImpl(sparkConf:
SparkConf)
.build()
}.getOrElse(executorContainer)
- new PodBuilder(executorPod)
+ val (maybeSecretsMountedPod, maybeSecretsMountedContainer) =
+ mountSecretsBootstrap.map { bootstrap =>
+ bootstrap.mountSecrets(executorPod, containerWithLimitCores)
--- End diff --
The submission client has to do more work in each step than the driver has
to do to configure executors. The submission client's steps also set up
accompanying Kubernetes resources and the driver's spark configuration, whereas
the driver only has to use those Kubernetes resources to mount volumes into the
pods.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
