Ngone51 commented on code in PR #38876:
URL: https://github.com/apache/spark/pull/38876#discussion_r1043027754
##########
core/src/main/scala/org/apache/spark/storage/BlockManager.scala:
##########
@@ -637,9 +637,11 @@ private[spark] class BlockManager(
def reregister(): Unit = {
// TODO: We might need to rate limit re-registering.
logInfo(s"BlockManager $blockManagerId re-registering with master")
- master.registerBlockManager(blockManagerId,
diskBlockManager.localDirsString, maxOnHeapMemory,
- maxOffHeapMemory, storageEndpoint)
- reportAllBlocks()
+ val id = master.registerBlockManager(blockManagerId,
diskBlockManager.localDirsString,
+ maxOnHeapMemory, maxOffHeapMemory, storageEndpoint, isReRegister = true)
+ if (id.executorId != BlockManagerId.INVALID_EXECUTOR_ID) {
+ reportAllBlocks()
+ }
Review Comment:
Good question. So ideally, a lost executor should be terminated in the end
anyways (whether killed by the driver or exit itself proactively)...if the lost
executor fails to terminate, there must be something wrong with it. So I think
terminate here won't make any difference to the result.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
