Github user sameeragarwal commented on a diff in the pull request:
https://github.com/apache/spark/pull/14412#discussion_r98605803
--- Diff:
core/src/main/scala/org/apache/spark/storage/BlockManagerMasterEndpoint.scala
---
@@ -188,24 +189,45 @@ class BlockManagerMasterEndpoint(
}
private def removeBlockManager(blockManagerId: BlockManagerId) {
+ val proactivelyReplicate =
conf.get("spark.storage.replication.proactive", "false").toBoolean
+
val info = blockManagerInfo(blockManagerId)
// Remove the block manager from blockManagerIdByExecutor.
blockManagerIdByExecutor -= blockManagerId.executorId
- // Remove it from blockManagerInfo and remove all the blocks.
- blockManagerInfo.remove(blockManagerId)
val iterator = info.blocks.keySet.iterator
while (iterator.hasNext) {
val blockId = iterator.next
val locations = blockLocations.get(blockId)
locations -= blockManagerId
if (locations.size == 0) {
blockLocations.remove(blockId)
+ logWarning(s"No more replicas available for $blockId !")
+ } else if ((blockId.isRDD || blockId.isInstanceOf[TestBlockId]) &&
proactivelyReplicate) {
+ // we only need to proactively replicate RDD blocks
+ // we also need to replicate this behavior for test blocks for
unit tests
+ // we send a message to a randomly chosen executor location to
replicate block
+ // assuming single executor failure, we find out how many replicas
existed before failure
+ val maxReplicas = locations.size + 1
--- End diff --
What happens if multiple executors are removed simultaneously? Depending on
the invocation sequence, is it possible for `maxReplicas` to be significantly
less than the original number of replicas?
---
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]
