[ https://issues.apache.org/jira/browse/SPARK-21425?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16095678#comment-16095678 ]
Ryan Williams commented on SPARK-21425: --------------------------------------- [~zsxwing] yea, it's static accumulators, and seems to only be a problem in "local" mode. I just updated [my comment above|https://issues.apache.org/jira/browse/SPARK-21425?focusedCommentId=16090017&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16090017] because I'd previously thought I'd been able to repro this in cluster mode, but now looking back at what I tried, that's not the case afaict. Small nit: I'm not sure what you mean by "local-cluster mode… will always be 0" though… I definitely get nonzero results on the driver with master "local\[*\]". Anyway, thanks for investigating. > LongAccumulator, DoubleAccumulator not threadsafe > ------------------------------------------------- > > Key: SPARK-21425 > URL: https://issues.apache.org/jira/browse/SPARK-21425 > Project: Spark > Issue Type: Bug > Components: Spark Core > Affects Versions: 2.2.0 > Reporter: Ryan Williams > Priority: Minor > > [AccumulatorV2 > docs|https://github.com/apache/spark/blob/v2.2.0/core/src/main/scala/org/apache/spark/util/AccumulatorV2.scala#L42-L43] > acknowledge that accumulators must be concurrent-read-safe, but afaict they > must also be concurrent-write-safe. > The same docs imply that {{Int}} and {{Long}} meet either/both of these > criteria, when afaict they do not. > Relatedly, the provided > [LongAccumulator|https://github.com/apache/spark/blob/v2.2.0/core/src/main/scala/org/apache/spark/util/AccumulatorV2.scala#L291] > and > [DoubleAccumulator|https://github.com/apache/spark/blob/v2.2.0/core/src/main/scala/org/apache/spark/util/AccumulatorV2.scala#L370] > are not thread-safe, and should be expected to behave undefinedly when > multiple concurrent tasks on the same executor write to them. > [Here is a repro repo|https://github.com/ryan-williams/spark-bugs/tree/accum] > with some simple applications that demonstrate incorrect results from > {{LongAccumulator}}'s. -- This message was sent by Atlassian JIRA (v6.4.14#64029) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org