Github user CodingCat commented on a diff in the pull request:
https://github.com/apache/spark/pull/96#discussion_r10387464
--- Diff: core/src/main/scala/org/apache/spark/storage/MemoryStore.scala ---
@@ -236,13 +236,23 @@ private class MemoryStore(blockManager: BlockManager,
maxMemory: Long)
while (maxMemory - (currentMemory - selectedMemory) < space &&
iterator.hasNext) {
val pair = iterator.next()
val blockId = pair.getKey
- if (rddToAdd.isDefined && rddToAdd == getRddId(blockId)) {
- logInfo("Will not store " + blockIdToAdd + " as it would
require dropping another " +
- "block from the same RDD")
- return false
+ // Apply the same-RDD rule for cache replacement. Quoted from the
+ // original RDD paper:
+ //
+ // When a new RDD partition is computed but there is not
enough
+ // space to store it, we evict a partition from the least
recently
+ // accessed RDD, unless this is the same RDD as the one with
the
+ // new partition. In that case, we keep the old partition in
memory
+ // to prevent cycling partitions from the same RDD in and out.
+ //
+ // TODO implement LRU eviction
+ rddToAdd match {
+ case Some(rddId) if rddId == getRddId(blockId) =>
+ // no-op
+ case _ =>
+ selectedBlocks += blockId
+ selectedMemory += pair.getValue.size
--- End diff --
Just a suggested alternative to LRU:
To minimize the number of affected RDDs, how about evicting the blocks from
those RDDs occupying the most memory space first, because in usual, all the
blocks in RDD are necessary for the computation, this approach may minimize the
chance for recomputation
---
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.
---
