[HIVE-24294] TezSessionPool sessions can throw AssertionError

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.0.0-alpha-1
• Labels: pull-request-available
• Assignee: Naresh P R
• Reporter: Naresh P R
• Created: 22/Oct/20 01:37
• Updated: 17/Nov/22 08:49
• Resolved: 29/Oct/20 19:48

Description

Whenever default TezSessionPool sessions are reopened for some reason, we are setting dagResources to null before close & setting it back in open
https://github.com/apache/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/exec/tez/TezSessionPoolManager.java#L498-L503
If there is an exception in sessionState.close(), we are not restoring the dagResource but moving the session back to TezSessionPool.

Below is sample stack trace where sessionState.close() fails:

2020-10-15T09:20:28,749 INFO  [HiveServer2-Background-Pool: Thread-25451]: client.TezClient (:()) - Failed to shutdown Tez Session via proxy
org.apache.tez.dag.api.SessionNotRunning: Application not running, applicationId=application_1602093123456_12345, yarnApplicationState=FINISHED, finalApplicationStatus=SUCCEEDED, trackingUrl=http://localhost:8088/proxy/application_1602093123456_12345/, diagnostics=Session timed out, lastDAGCompletionTime=1602997683786 ms, sessionTimeoutInterval=600000 ms
Session stats:submittedDAGs=2, successfulDAGs=2, failedDAGs=0, killedDAGs=0        at org.apache.tez.client.TezClientUtils.getAMProxy(TezClientUtils.java:910) 
        at org.apache.tez.client.TezClient.getAMProxy(TezClient.java:1060) 
        at org.apache.tez.client.TezClient.stop(TezClient.java:743) 
        at org.apache.hadoop.hive.ql.exec.tez.TezSessionState.closeClient(TezSessionState.java:789) 
        at org.apache.hadoop.hive.ql.exec.tez.TezSessionState.close(TezSessionState.java:756) 
        at org.apache.hadoop.hive.ql.exec.tez.TezSessionPoolSession.close(TezSessionPoolSession.java:111) 
        at org.apache.hadoop.hive.ql.exec.tez.TezSessionPoolManager.reopenInternal(TezSessionPoolManager.java:496) 
        at org.apache.hadoop.hive.ql.exec.tez.TezSessionPoolManager.reopen(TezSessionPoolManager.java:487) 
        at org.apache.hadoop.hive.ql.exec.tez.TezSessionPoolSession.reopen(TezSessionPoolSession.java:228) 
        at org.apache.hadoop.hive.ql.exec.tez.TezTask.getNewTezSessionOnError(TezTask.java:531) 
        at org.apache.hadoop.hive.ql.exec.tez.TezTask.submit(TezTask.java:546) 
        at org.apache.hadoop.hive.ql.exec.tez.TezTask.execute(TezTask.java:221)

Because of this, all new queries using this corrupted sessions are failing with below exception

Caused by: java.lang.AssertionError: Ensure called on an unitialized (or closed) session 41774265-b7da-4d58-84a8-1bedfd597aecCaused by: java.lang.AssertionError: Ensure called on an unitialized (or closed) session 41774265-b7da-4d58-84a8-1bedfd597aec at org.apache.hadoop.hive.ql.exec.tez.TezSessionState.ensureLocalResources(TezSessionState.java:685)

Issue Links

• links to: GitHub Pull Request #1596