Github user pwendell commented on a diff in the pull request:
https://github.com/apache/spark/pull/7770#discussion_r36055835
--- Diff: core/src/main/scala/org/apache/spark/ui/ToolTips.scala ---
@@ -62,6 +62,13 @@ private[spark] object ToolTips {
"""Time that the executor spent paused for Java garbage collection
while the task was
running."""
+ val PEAK_EXECUTION_MEMORY =
+ """Execution memory refers to the memory used by internal data
structures created during
+ shuffles, aggregations and joins when Tungsten is enabled. The
value of this accumulator
+ should be approximately the sum of the peak sizes across all such
data structures created
+ in this task. For SQL jobs, this only tracks all unsafe operators,
broadcast joins, and
+ external sort."""
--- End diff --
Yeah, I agree the inconsistency is the biggest problem. I actually think
the best thing might be to rename "spark.shuffle.memoryFraction" to be
"spark.execution.memoryFraction" so it's consistent with this header (we can
still accept the old one for backwards compatibility).
Overall I feel that "execution" is a substantially better name than shuffle
for this. The thing is, you need a slightly generic term here because it is a
catch all for several different types of memory used at runtime (concerning
joins, aggregations, and other buffers). It can be defined in more detail in
the tooltip.
---
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]
