josh0yeh commented on a change in pull request #9587: [FLINK-13685][yarn]
Validate YARN deployment against YARN scheduler maximum vcore size
URL: https://github.com/apache/flink/pull/9587#discussion_r321978438
##########
File path:
flink-yarn/src/main/java/org/apache/flink/yarn/AbstractYarnClusterDescriptor.java
##########
@@ -483,12 +483,14 @@ private void
validateClusterSpecification(ClusterSpecification clusterSpecificat
}
final int yarnMinAllocationMB =
yarnConfiguration.getInt(YarnConfiguration.RM_SCHEDULER_MINIMUM_ALLOCATION_MB,
0);
+ final int yarnSchedulerMaxVcores =
yarnConfiguration.getInt(YarnConfiguration.RM_SCHEDULER_MAXIMUM_ALLOCATION_VCORES,
0);
Review comment:
@wangyang0918 thanks for your feedback. Please correct me if I am wrong.
AFAIK, maximumResourceCapability.getVirtualCores() seems to get the aggregated
amount of cores from YARN node managers that are across the YARN cluster. But
the value of yarn.scheduler.maximum-allocation-vcores in yarn-site.xml is still
governing the maximum of vcores per container that flink can get from YARN.
In other words, if maximum-allocation-vcores is capped at 4, even the total
cores in yarn cluster is plenty. Flink can only get container with no more than
4 cores.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
