xintongsong commented on a change in pull request #11320: 
[FLINK-16437][runtime] Make SlotManager allocate resource from ResourceManager 
at the worker granularity.
URL: https://github.com/apache/flink/pull/11320#discussion_r388861967
 
 

 ##########
 File path: 
flink-yarn/src/main/java/org/apache/flink/yarn/entrypoint/YarnResourceManagerFactory.java
 ##########
 @@ -81,4 +91,27 @@ public static YarnResourceManagerFactory getInstance() {
                        webInterfaceUrl,
                        resourceManagerMetricGroup);
        }
+
+       @Override
+       protected CPUResource getDefaultCpus(final Configuration configuration) 
{
+               int fallback = 
configuration.getInteger(YarnConfigOptions.VCORES);
+               double cpuCoresDouble = 
TaskExecutorProcessUtils.getCpuCoresWithFallback(configuration, 
fallback).getValue().doubleValue();
+               @SuppressWarnings("NumericCastThatLosesPrecision")
+               long cpuCoresLong = Math.max((long) Math.ceil(cpuCoresDouble), 
1L);
+               //noinspection FloatingPointEquality
+               if (cpuCoresLong != cpuCoresDouble) {
+                       LOG.info(
+                               "The amount of cpu cores must be a positive 
integer on Yarn. Rounding {} up to the closest positive integer {}.",
+                               cpuCoresDouble,
+                               cpuCoresLong);
+               }
+               if (cpuCoresLong > Integer.MAX_VALUE) {
 
 Review comment:
   I think it is not robust to make the correctness of 
`YarnResourceManagerFactory` depending on another class 
`YarnClusterDescriptor`. There is no harm for double checking it, right?

----------------------------------------------------------------
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

Reply via email to