Github user squito commented on a diff in the pull request:
https://github.com/apache/spark/pull/14617#discussion_r109015878
--- Diff:
core/src/main/scala/org/apache/spark/storage/StorageStatusListener.scala ---
@@ -74,8 +74,11 @@ class StorageStatusListener(conf: SparkConf) extends
SparkListener {
synchronized {
val blockManagerId = blockManagerAdded.blockManagerId
val executorId = blockManagerId.executorId
- val maxMem = blockManagerAdded.maxMem
- val storageStatus = new StorageStatus(blockManagerId, maxMem)
+ // This two fields are compatible with old event logs, in which
there only has max on heap
+ // memory in the event log. So maxOnHeapMem will use maxMem,
maxOffHeapMem will set to 0.
+ val maxOnHeapMem =
blockManagerAdded.maxOnHeapMem.getOrElse(blockManagerAdded.maxMem)
--- End diff --
actually wait ... are you sure about this? it looks to me like in old
event logs, we actually dont' know the breakdown between onheap and offheap.
I'm pretty sure the value in old event logs starts from here:
https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/storage/BlockManager.scala#L154
so then we have to decide how to represent that old info. The options are:
a) don't show anything, with an empty string in the display, and the types
are `Option` even in the rest api
b) show a value of `0` for both of them if they are missing
c) assume all the memory was onheap
this is implementing (c). I'd prefer (a), if my understanding is correct.
---
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]
