[YARN-6640] AM heartbeat stuck when responseId overflows MAX_INT

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 2.9.0, 3.0.0-beta1, 2.8.2
• Target Version/s: 2.9.0, 3.0.0-beta1, 2.8.2
• Hadoop Flags: Reviewed
• Assignee: Botong Huang Assign to me
• Reporter: Botong Huang
• Created: 24/May/17 17:41
• Updated: 04/Oct/17 09:11
• Resolved: 25/Aug/17 15:21

Description

The current code in ApplicationMasterService:
if ((request.getResponseId() + 1) == lastResponse.getResponseId())

{/* old heartbeat */ return lastResponse;}

else if (request.getResponseId() + 1 < lastResponse.getResponseId())

{ throw ... }

process the heartbeat...

When a heartbeat comes in, in usual case we are expecting request.getResponseId() == lastResponse.getResponseId(). The “if“ is for the duplicate heartbeat that’s one step old, the “else if” is to throw and complain for heartbeats more than two steps old, otherwise we accept the new heartbeat and process it.

So the bug is: when lastResponse.getResponseId() == MAX_INT, the newest heartbeat comes in with responseId == MAX_INT. However reponseId + 1 will be MIN_INT, and we will fall into the “else if” case and RM will throw. Then we are stuck here…

Issue Links

• breaks: YARN-7211 AMSimulator in SLS does't work due to refactor of responseId
• breaks: YARN-7112 TestAMRMProxy is failing with invalid request
• breaks: YARN-7122 TestReservationSystemInvariants and TestSLSRunner fail