Skip to content

[SPARK-23508][CORE] Fix BlockmanagerId in case blockManagerIdCache cause oom #20667

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
caneGuy wants to merge 11 commits into from
Closed

[SPARK-23508][CORE] Fix BlockmanagerId in case blockManagerIdCache cause oom #20667

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
fc1b6a0
Use timeStampedHashMap for BlockmanagerId in case blockManagerIdCache…
caneGuy Feb 24, 2018
59f34e1
Revert "Use timeStampedHashMap for BlockmanagerId in case blockManage…
caneGuy Feb 26, 2018
e5e3eff
Use softreference instead
caneGuy Feb 26, 2018
c741b58
Remove unused import
caneGuy Feb 26, 2018
36d66e1
Revert "Remove unused import"
caneGuy Feb 26, 2018
e5b793d
Revert "Use softreference instead"
caneGuy Feb 26, 2018
8f3cd47
Use guava cache instead
caneGuy Feb 26, 2018
dc522fc
Update
caneGuy Feb 27, 2018
b28985b
Refine comment
caneGuy Feb 27, 2018
3379899
Update documentation
caneGuy Feb 28, 2018
bf79f4d
Refine comment
caneGuy Feb 28, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 11 additions & 5 deletions core/src/main/scala/org/apache/spark/storage/BlockManagerId.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ import java.util.concurrent.ConcurrentHashMap

import org.apache.spark.SparkContext
import org.apache.spark.annotation.DeveloperApi
import org.apache.spark.util.Utils
import org.apache.spark.util.{TimeStampedHashMap, Utils}

/**
* :: DeveloperApi ::
Expand Down Expand Up @@ -123,7 +123,8 @@ private[spark] object BlockManagerId {
execId: String,
host: String,
port: Int,
topologyInfo: Option[String] = None): BlockManagerId =
topologyInfo: Option[String] = None,
clearOldValues: Boolean = false): BlockManagerId =
getCachedBlockManagerId(new BlockManagerId(execId, host, port, topologyInfo))

def apply(in: ObjectInput): BlockManagerId = {
Expand All @@ -132,10 +133,15 @@ private[spark] object BlockManagerId {
getCachedBlockManagerId(obj)
}

val blockManagerIdCache = new ConcurrentHashMap[BlockManagerId, BlockManagerId]()
val blockManagerIdCache = new TimeStampedHashMap[BlockManagerId, BlockManagerId](true)
Copy link
Contributor

@jiangxb1987 jiangxb1987 Feb 25, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will this cause serious performance regression? TimeStampedHashMap will copy all the old values on update. cc @cloud-fan


def getCachedBlockManagerId(id: BlockManagerId): BlockManagerId = {
def getCachedBlockManagerId(id: BlockManagerId, clearOldValues: Boolean = false): BlockManagerId =
{
blockManagerIdCache.putIfAbsent(id, id)
blockManagerIdCache.get(id)
val blockManagerId = blockManagerIdCache.get(id)
if (clearOldValues) {
blockManagerIdCache.clearOldValues(System.currentTimeMillis - Utils.timeStringAsMs("10d"))
Copy link
Member

@Ngone51 Ngone51 Feb 24, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

10 days? I don't think time can be a judging criteria to decide whether we should remove a cached id or not, even if you set the time threshold far less/greater than '10d'. Think about a extreamly case that a block could be frequently got all the time during the app‘s running. So, it would be certainly removed from cache due to the time threshold, and recached next time we get it, and repeatedly.

Copy link
Contributor Author

@caneGuy caneGuy Feb 26, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@Ngone51 Thanks.i also though about remove when we delete a block.
In this case, it is history replaying which will trigger this problem,and we do not delete any block actually.
Maybe use weakreference better as @jiangxb1987 mentioned?WDYT?
Thanks again!

Copy link
Contributor

@cloud-fan cloud-fan Feb 26, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we simply use com.google.common.cache.Cache? which has a size limitation and we don't need to worry about OOM.

}
blockManagerId.get
}
}
2 changes: 1 addition & 1 deletion core/src/main/scala/org/apache/spark/util/JsonProtocol.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -946,7 +946,7 @@ private[spark] object JsonProtocol {
val executorId = (json \ "Executor ID").extract[String].intern()
val host = (json \ "Host").extract[String].intern()
val port = (json \ "Port").extract[Int]
BlockManagerId(executorId, host, port)
BlockManagerId(executorId, host, port, clearOldValues = true)
}

private object JOB_RESULT_FORMATTED_CLASS_NAMES {
Expand Down