[YARN-7270] Fix unsafe casting from long to int for class Resource and its sub-classes

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.1.0
• Fix Version/s: 2.9.0, 3.0.0, 3.1.0
• Component/s: resourcemanager
• Hadoop Flags: Reviewed
• Assignee: Yufei Gu
• Reporter: Yufei Gu
• Created: 29/Sep/17 00:05
• Updated: 28/Jun/18 09:15
• Resolved: 13/Oct/17 19:44

Description

Class Resource has three sub classes(FixedValueResource, LightWeightResource, and ResourcePBImpl). Only FixedValueResource handle long-to-int casting nicely. The other two didn't. This bug is introduced by resource type feature and causes several unit test failures. For example:

Error Message

expected:<<memory:4096, vCores:4>> but was:<<memory:4096, vCores:-1>>
Stacktrace

java.lang.AssertionError: expected:<<memory:4096, vCores:4>> but was:<<memory:4096, vCores:-1>>
	at org.junit.Assert.fail(Assert.java:88)
	at org.junit.Assert.failNotEquals(Assert.java:743)
	at org.junit.Assert.assertEquals(Assert.java:118)
	at org.junit.Assert.assertEquals(Assert.java:144)
	at org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.TestFSAppAttempt.testHeadroomWithBlackListedNodes(TestFSAppAttempt.java:325)

Issue Links

• Blocked: YARN-7310 TestAMRMProxy#testAMRMProxyE2E fails with FairScheduler
• duplicates: YARN-7314 Multiple Resource manager test cases failing
• is caused by: YARN-4844 Add getMemorySize/getVirtualCoresSize to o.a.h.y.api.records.Resource
• is related to: YARN-7314 Multiple Resource manager test cases failing