prakharjain09 commented on a change in pull request #27539: [SPARK-30786]
[CORE] Fix Block replication failure propogation issue in BlockManager
URL: https://github.com/apache/spark/pull/27539#discussion_r377524849
##########
File path:
core/src/main/scala/org/apache/spark/network/netty/NettyBlockRpcServer.scala
##########
@@ -105,7 +105,11 @@ class NettyBlockRpcServer(
val blockId = BlockId(uploadBlock.blockId)
logDebug(s"Receiving replicated block $blockId with level ${level} " +
s"from ${client.getSocketAddress}")
- blockManager.putBlockData(blockId, data, level, classTag)
+ val blockStored = blockManager.putBlockData(blockId, data, level,
classTag)
+ if (!blockStored) {
+ throw new Exception(s"Upload block for $blockId failed. This " +
+ s"mostly happens when there is not sufficient space available to
store the block.")
+ }
Review comment:
Yes I initially used responseContext.ononFailure(e) only. The issue is that
responseContext variable is not available in the streaming flow
https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/storage/BlockManager.scala#L669
. Also passing responseContext variable till that point will need interface
changes in BlockDataManager#putBlockDataAsStream interface method. So we have
to use exception in that flow. To make things consistent at both the places, I
used exception here also. @Ngone51 Should we use responseContext.onFailure
instead of exception here?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
