mridulm commented on a change in pull request #31876:
URL: https://github.com/apache/spark/pull/31876#discussion_r597391662
##########
File path: core/src/main/scala/org/apache/spark/scheduler/MapStatus.scala
##########
@@ -216,7 +223,10 @@ private[spark] class HighlyCompressedMapStatus private (
}
override def readExternal(in: ObjectInput): Unit = Utils.tryOrIOException {
- loc = BlockManagerId(in)
+ // TODO(wuyi): config
+ val location = "org.apache.spark.storage.BlockManagerId"
+ loc = Utils.classForName(location).newInstance().asInstanceOf[Location]
+ loc.readExternal(in)
Review comment:
When this is fixed, do the same for `CompressedMapStatus` as well.
##########
File path: core/src/main/scala/org/apache/spark/scheduler/MapStatus.scala
##########
@@ -28,16 +28,23 @@ import org.apache.spark.internal.config
import org.apache.spark.storage.BlockManagerId
import org.apache.spark.util.Utils
+trait Location extends Externalizable {
Review comment:
I might have missed some context here, but why `Externalizable` and not
`Serializable` ?
##########
File path: core/src/main/scala/org/apache/spark/storage/BlockManagerId.scala
##########
@@ -129,7 +132,7 @@ private[spark] object BlockManagerId {
def apply(in: ObjectInput): BlockManagerId = {
val obj = new BlockManagerId()
obj.readExternal(in)
- getCachedBlockManagerId(obj)
+ obj
Review comment:
Why would we not cache this ?
##########
File path: core/src/main/scala/org/apache/spark/scheduler/MapStatus.scala
##########
@@ -28,16 +28,23 @@ import org.apache.spark.internal.config
import org.apache.spark.storage.BlockManagerId
import org.apache.spark.util.Utils
+trait Location extends Externalizable {
+ def host: String
+ def port: Int
+ def hostPort: String
+ def executorId: String = "unknown"
Review comment:
Based on discussion, if we are going to mirror everything in
`BlockManagerId`, why not expose that directly with ability to subclass it and
avoid `Location` altogether ?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
