[OOZIE-1394] Fix Bugs in Job and SLA Events

• Type: Sub-task
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: trunk
• Fix Version/s: 4.0.0
• Assignee: Mona Chitnis
• Reporter: Mona Chitnis
• Created: 29/May/13 23:11
• Updated: 31/Aug/13 00:02
• Resolved: 11/Jun/13 04:41

Description

This JIRA fixes following bugs related to Job and SLA Events:

(1) no parent id in WORKFLOW_JOB STARTED event. the workflow is started by
coord action.
the message text:

{"status":"RUNNING","id":"0000134-130528180458682-oozie-oozi-W","startTime":1369860763055}

(2) no COORDINATOR_ACTION STARTED event, after resume of suspended action.

(3) no nominal time in suspended COORDINATOR_ACTION event.
the message text:

{"status":"SUSPENDED","id":"0000135-130528180458682-oozie-oozi-C@1","parentId":"0000135-130528180458682-oozie-oozi-C"}

(4) There is no msg=JOB START event.
Job start time = Thu, 30 May 2013 20:19:39 GMT
$ zgrep msgType=JOB
/home/y/libexec/yjava_tomcat/logs/qe/oozie.log-2013-05-30-20.gz | grep 000001
This shows SUCCESS event only.

(5) submit the same job again, 0000003-130530200622405-oozie-oozi-W.
there is no event at all.
$ zgrep msgType=SLA
/home/y/libexec/yjava_tomcat/logs/qe/oozie.log-2013-05-30-22.gz
$ zgrep msgType=JOB
/home/y/libexec/yjava_tomcat/logs/qe/oozie.log-2013-05-30-22.gz

(6) submit another job, 0000004-130530200622405-oozie-oozi-W.
then sla/job messages of previous jobs show up:
$ zgrep msgType=SLA
/home/y/libexec/yjava_tomcat/logs/qe/oozie.log-2013-05-30-23.gz
$ zgrep msgType=JOB
/home/y/libexec/yjava_tomcat/logs/qe/oozie.log-2013-05-30-23.gz

(7) when actual start is very close to expected start, it might happen that job event is processed and received much later causing END_MISS when it could be END_MET

Issue Links

• requires: OOZIE-1418 Fix bugs around ActionKillX not setting end time, V2SLAServlet and exception handling for event threads