Github user pwendell commented on a diff in the pull request:
https://github.com/apache/spark/pull/4067#discussion_r23951392
--- Diff: core/src/main/scala/org/apache/spark/CacheManager.scala ---
@@ -47,9 +49,13 @@ private[spark] class CacheManager(blockManager:
BlockManager) extends Logging {
val inputMetrics = blockResult.inputMetrics
val existingMetrics = context.taskMetrics
.getInputMetricsForReadMethod(inputMetrics.readMethod)
- existingMetrics.addBytesRead(inputMetrics.bytesRead)
+ existingMetrics.incBytesRead(inputMetrics.bytesRead)
- new InterruptibleIterator(context,
blockResult.data.asInstanceOf[Iterator[T]])
+ val iter = blockResult.data.asInstanceOf[Iterator[T]]
+ new InterruptibleIterator(context,
AfterNextInterceptingIterator(iter, (next: T) => {
+ existingMetrics.incRecordsRead(1)
--- End diff --
The question is - how expensive is the thing we are doing inside of the
override method? In those other cases I think we're just checking the value of
a single variable that doesn't change often (i.e. checking for interrupted). In
the past we've seen performance regressions from anything more expensive than
this: See
https://github.com/apache/spark/commit/f708dda81ed5004325591fcc31cd79a8afa580db.
The cost of CAS is hardware dependent, but can be expensive on machines
with large numbers of cores because in many case there is a single shared bus
lock. Volatile is similar, basically if you think about it maybe you have 16
cores and they each need to constantly invalidate each-other's local copy of
the variable.
---
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]
