[
https://issues.apache.org/jira/browse/HIVE-17292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16127268#comment-16127268
]
Peter Vary commented on HIVE-17292:
-----------------------------------
Actually checked the configuration of the standalone mode, and here is what I
have found:
{code:title=data/conf/spark/standalone/hive-site.xml}
<property>
<name>spark.master</name>
<value>local-cluster[2,2,1024]</value>
</property>
{code}
This suggests, that we planned to use 2 executors with 2 cores each, but the
{{SparkSessionImpl.getMemoryAndCores}} overrides this with configuration values
we do not set in the conf.
{code:title=SparkSessionImpl}
@Override
public ObjectPair<Long, Integer> getMemoryAndCores() throws Exception {
[..]
if (masterURL.startsWith("spark")) {
[..]
} else {
int coresPerExecutor = sparkConf.getInt("spark.executor.cores", 1);
totalCores = numExecutors * coresPerExecutor;
}
[..]
}
{code}
Based on our conversation here and on HIVE-17291, I think the ideal solution
would be to use the same logic for the "local.\*" master, as for the
"spark.\*". This means that we have to regenerate many golden files, since we
will change the result of the explains in the same way as for the
TestMiniSparkOnYarnCliDriver.
If you agree [~lirui], I could provide a patch for it.
Thanks,
Peter
> Change TestMiniSparkOnYarnCliDriver test configuration to use the configured
> cores
> ----------------------------------------------------------------------------------
>
> Key: HIVE-17292
> URL: https://issues.apache.org/jira/browse/HIVE-17292
> Project: Hive
> Issue Type: Sub-task
> Components: Spark, Test
> Affects Versions: 3.0.0
> Reporter: Peter Vary
> Assignee: Peter Vary
> Attachments: HIVE-17292.1.patch, HIVE-17292.2.patch,
> HIVE-17292.3.patch
>
>
> Currently the {{hive-site.xml}} for the {{TestMiniSparkOnYarnCliDriver}} test
> defines 2 cores, and 2 executors, but only 1 is used, because the MiniCluster
> does not allows the creation of the 3rd container.
> The FairScheduler uses 1GB increments for memory, but the containers would
> like to use only 512MB. We should change the fairscheduler configuration to
> use only the requested 512MB
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
