[
https://issues.apache.org/jira/browse/OOZIE-3439?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16773953#comment-16773953
]
Julia Kinga Marton commented on OOZIE-3439:
-------------------------------------------
[~shubham.chhabra], you said that in case of Hive1 jobs the child IDs are
populated. I have checked the HiveMain code, and the patterns used for this job
is the following one:
{code:java}
static final Pattern[] HIVE_JOB_IDS_PATTERNS = {
Pattern.compile("Ended Job = (job_\\S*)"),
Pattern.compile("Submitted application (application[0-9_]*)")
};{code}
This pattern does not contain the pattern you included in Hive2Main. If this
pattern is missing it means that there is line in log what will match one of
the patterns above. Not?
> Hive2 action is not parsing application ID for TEZ from log file properly
> -------------------------------------------------------------------------
>
> Key: OOZIE-3439
> URL: https://issues.apache.org/jira/browse/OOZIE-3439
> Project: Oozie
> Issue Type: Bug
> Components: action
> Affects Versions: trunk
> Reporter: Shubham
> Assignee: Shubham
> Priority: Major
> Attachments: OOZIE-3439-001.patch
>
>
> Oozie workflow does not populate ChildJobUrl for Hive2 Action while Hive1 is
> able to find child job ids.
> I looked at the code and found that pattern is not correct for hive2 action
> logs generated in usercache.
> {code:java}
> static final Pattern[] HIVE2_JOB_IDS_PATTERNS = {
> Pattern.compile("Ended Job = (job_\\S*)"),
> Pattern.compile("Submitted application (application[0-9_]*)"),
> Pattern.compile("Running with YARN Application = (application[0-9_]*)")
> }
> {code}
> Adding below pattern should help in getting Hive 2 action Tez application id
> {code:java}
> Pattern.compile("Executing on YARN cluster with App id (application[0-9_]*)"),
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
