Github user squito commented on a diff in the pull request:
https://github.com/apache/spark/pull/6648#discussion_r32023078
--- Diff: core/src/main/scala/org/apache/spark/scheduler/MapStatus.scala ---
@@ -32,6 +32,9 @@ private[spark] sealed trait MapStatus {
/** Location where this task was run. */
def location: BlockManagerId
+ /** stage attempt for the ShuffleMapTask */
+ def stageAttemptId: Int
--- End diff --
`MapStatus` is now needs to keep a `stageAttemptId` so the fetch side knows
which stage attempt to read data from. This is a negligible increase to the
size of the map status when we have just one stage attempt, since its just an
extra int. But when there are multiple attempts, we now have one `MapStatus`
per (location, stageAttempt). In pathalogical cases, this will result in many
more `MapStatus`.
I needed to update the serialization of `MapStatus`, but I believe this
should be covered by existing tests.
---
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]
