Github user squito commented on a diff in the pull request:
https://github.com/apache/spark/pull/2851#discussion_r24642955
--- Diff: core/src/main/scala/org/apache/spark/storage/StorageUtils.scala
---
@@ -118,8 +140,20 @@ class StorageStatus(val blockManagerId:
BlockManagerId, val maxMem: Long) {
} else {
None
}
+ case BroadcastBlockId(broadcastId, _) =>
+ // Actually remove the block, if it exists
+ if (_broadcastBlocks.contains(broadcastId)) {
+ val removed = _broadcastBlocks(broadcastId).remove(blockId)
+ // If the given RDD has no more blocks left, remove the RDD
--- End diff --
actually now I am thinking there is a lot of copy-pasting that could be
cleaned up. you could mostly merge this w/ the block above if you did
something like:
```
case rddOrBlockId @ (_: BroadcastBlockId | _: RDDBlockId) =>
val (id, blockMap) = getIdAndBlockMap(rddOrBlockId)
val removed = blockMap(id).remove(blockId)
...
```
where the helper function `getIdAndBlockMap` is something like:
```
def getIdAndBlockMap(blockId: BlockId): (Int, Map[BlockId, BlockStatus]) =
blockId match {
case RDDBlockId(rddId, _) => (rddId, _rddBlocks)
case BroadcastBlockId(broadcastId, _) => (broadcastId, _broadcastBlocks)
}
```
and then you could do a similar thing in a few other places. You could
also take this a step further, and even merge `_rddBlocks` and
`_broadcastBlocks` into a `EnumMap[BlockType, Map[BlockId, BlockStatus]]` if
you made a new `public enum BlockType{ RDD,Broadcast}`, but that might not
really help much since at the end of the day you do want separate getter
methods for the RDD and Broadcast stuff
---
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]
