You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Yunling Cai (JIRA)" <ji...@apache.org> on 2018/08/27 16:42:00 UTC
[jira] [Commented] (SPARK-25091) UNCACHE TABLE, CLEAR CACHE,
rdd.unpersist() does not clean up executor memory
[ https://issues.apache.org/jira/browse/SPARK-25091?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16593919#comment-16593919 ]
Yunling Cai commented on SPARK-25091:
-------------------------------------
Thanks [~Chao Fang] for working on this! I have changed the ticket title.
Quick question: does this mean this is just a UI issue where executor information was shown incorrectly? As we saw the cache tables start falling onto disk even though we have uncache the previous copy of it. We also started seeing duplicate entries on the storage tab for same table and this is why we think the memory clean up may have actual problems.
Steps to reproduce:
CACHE TABLE A
UNCACHE TABLE A
CACHE TABLE A
REFRESH TABLE has a similar behavior.
Thanks!
> UNCACHE TABLE, CLEAR CACHE, rdd.unpersist() does not clean up executor memory
> -----------------------------------------------------------------------------
>
> Key: SPARK-25091
> URL: https://issues.apache.org/jira/browse/SPARK-25091
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 2.3.1
> Reporter: Yunling Cai
> Priority: Critical
>
> UNCACHE TABLE and CLEAR CACHE does not clean up executor memory.
> Through Spark UI, although in Storage, we see the cached table removed. In Executor, the executors continue to hold the RDD and the memory is not cleared. This results in huge waste in executor memory usage. As we call CACHE TABLE, we run into issues where the cached tables are spilled to disk instead of reclaiming the memory storage.
> Steps to reproduce:
> CACHE TABLE test.test_cache;
> UNCACHE TABLE test.test_cache;
> == Storage shows table is not cached; Executor shows the executor storage memory does not change ==
> CACHE TABLE test.test_cache;
> CLEAR CACHE;
> == Storage shows table is not cached; Executor shows the executor storage memory does not change ==
> Similar behavior when using pyspark df.unpersist().
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org