Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/5669#discussion_r29004054
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/tree/GradientBoostedTrees.scala ---
@@ -177,102 +177,108 @@ object GradientBoostedTrees extends Logging {
treeStrategy.assertValid()
// Cache input
- if (input.getStorageLevel == StorageLevel.NONE) {
+ val persistedInput = if (input.getStorageLevel == StorageLevel.NONE) {
input.persist(StorageLevel.MEMORY_AND_DISK)
- }
+ true
+ } else false
- timer.stop("init")
+ try {
--- End diff --
I too think it makes some sense but have not seen it done elsewhere. I
think I asked about it when I first started working on Spark and was told
something like @srowen said. Even if it's "leaked," other RDDs can still push
it out of memory or disk if needed, right? Have you encountered cases where
that does not happen?
---
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]
