[REEF-80] ID mismatch bug in YarnContainerManager#appendByDeleteAndCreate

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.11
• Component/s: REEF-Runtime-YARN, REEF-Tests
• Assignee: Yingda Chen
• Reporter: John Yang
• Created: 25/Dec/14 14:59
• Updated: 16/Feb/15 05:06
• Resolved: 10/Feb/15 09:00

Description

One integration test(reef-job-TEST_CloseEvaluatorTest) fails on YARN due to an ID mismatch bug in YarnContainerManager#appendByDeleteAndCreate. This bug also appeared in a REEF application.

Below is the stacktrace of the failed test. Please find attached the full log. One should also be able to reproduce this by running the tests on YARN as instructed at https://github.com/apache/incubator-reef/pull/38.

Caused by: org.apache.hadoop.yarn.exceptions.YarnRuntimeException: java.lang.RuntimeException: Cannot append by read-append-delete-create with exception.
Caused by: java.lang.RuntimeException: Cannot append by read-append-delete-create with exception.
Caused by: org.apache.hadoop.ipc.RemoteException(java.io.FileNotFoundException): ID mismatch. Request id and saved id: 165643 , 165644 for file /ReefApplications/application_1418954976955_0112/evaluatorsChangesLog

Issue Links

• blocks: REEF-147 Release 0.11
• links to: PR #43 with a hotfix for this
• links to: PR-44