prakharjain09 commented on a change in pull request #27864: [SPARK-20732][CORE]
Decommission cache blocks to other executors when an executor is decommissioned
URL: https://github.com/apache/spark/pull/27864#discussion_r399773819
##########
File path: core/src/main/scala/org/apache/spark/storage/BlockManager.scala
##########
@@ -1560,26 +1567,30 @@ private[spark] class BlockManager(
def replicateBlock(
blockId: BlockId,
existingReplicas: Set[BlockManagerId],
- maxReplicas: Int): Unit = {
+ maxReplicas: Int,
+ forceFetch: Boolean = true,
Review comment:
Parameter name changes to forceFetchPeers.
To decommision this block managers, we have to replicate all the cache
blocks to some other peers. To replicate n blocks, we are going to call this
method n times. This method force fetches the peers each time it is called. So
it will end up making n getPeer calls which are not required. So we have made
it configurable (by default to true). The decommission flow refreshes the peer
list once in beginning and then calls replicateBlock(forceFetchPeers=false) to
avoid n calls.
----------------------------------------------------------------
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]
