holdenk commented on a change in pull request #28370:
URL: https://github.com/apache/spark/pull/28370#discussion_r426338033
##########
File path: core/src/main/scala/org/apache/spark/storage/BlockManager.scala
##########
@@ -1829,7 +1901,58 @@ private[spark] class BlockManager(
data.dispose()
}
+ /**
+ * Class to handle block manager decommissioning retries
+ * It creates a Thread to retry offloading all RDD cache blocks
+ */
+ private class BlockManagerDecommissionManager(conf: SparkConf) {
Review comment:
So there are two conversations I want to have about this with you
@Ngone51 now to make sure I'm understanding what you're trying to express.
There already is a second follow up PR that extends the
`BlockManagerDecommissionManager` already exists, so I'm not sure I agree with
your reasoning. If it was only just for some possible future implementation
that didn't already exist I'd be more inclined to simplify. Maybe you can take
a look at https://issues.apache.org/jira/browse/SPARK-20624 ,
https://issues.apache.org/jira/browse/SPARK-20629 and
https://github.com/apache/spark/pull/28331 for context.
I want to understand your -1 here because that has some pretty strong
meanings in the context of a code change. A -1 is generally viewed as
expressing a veto, which I don't believe you have in the project (of course I
was out for a month in the hospital last year so if you do please let point me
to thread). Even if you don't have a veto in the project is it your intention
to say that if you did have a veto you would block this code change? A veto is
generally a very strong expression, and I'm worried I'm not understanding your
reasoning since this seems like a relatively minor issue.
----------------------------------------------------------------
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
