Github user zsxwing commented on a diff in the pull request:
https://github.com/apache/spark/pull/8710#discussion_r39337844
--- Diff: core/src/main/scala/org/apache/spark/SparkContext.scala ---
@@ -348,10 +348,24 @@ class SparkContext(config: SparkConf) extends Logging
with ExecutorAllocationCli
// Thread Local variable that can be used by users to pass information
down the stack
private val localProperties = new InheritableThreadLocal[Properties] {
- override protected def childValue(parent: Properties): Properties =
new Properties(parent)
+ override protected def childValue(parent: Properties): Properties = {
+ // Note: make a clone such that changes in the parent properties
aren't reflected in
+ // the those of the children threads, which has confusing semantics
(SPARK-10564).
+ val p = new Properties
+ val filtered = parent.asScala.filter { case (k, _) =>
+ !nonInheritedLocalProperties.contains(k)
+ }
+ p.putAll(filtered.asJava)
+ p
+ }
override protected def initialValue(): Properties = new Properties()
}
+ /**
+ * Keys of local properties that should not be inherited by children
threads.
+ */
+ private[spark] val nonInheritedLocalProperties: HashSet[String] = new
HashSet[String]
--- End diff --
Exposing a mutable `HashSet` in the thread-safe `SparkContext` looks
dangerous. Actually, I suggest we don't need to add
`nonInheritedLocalProperties` in the master branch. How about just cloning the
parent properties without adding the `nonInheritedLocalProperties` logic?
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
