[
https://issues.apache.org/jira/browse/HIVE-18672?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16363361#comment-16363361
]
Sahil Takiar commented on HIVE-18672:
-------------------------------------
Attaching updated patch:
* Realized that when {{JobHandle#State = QUEUED}} that Spark job id isn't known
yet, and always returns -1; so I removed it
* Changed {{SparkJobMonitor}} to create a new {{LogHelper}} rather than use the
default one in {{SessionState}}; this makes the Hive logs much clearer;
anything printed from {{SparkJobMonitor}} will now be displayed in the logs as
coming from {{SparkJobMonitor}} rather than {{SessionState}}
* Added Spark job id to logs printed when in the {{CANCELLED}} state (added in
HIVE-18671)
* Added Spark job to the logs printed when in the {{SUCCEEDED}} state
> Printed state in RemoteSparkJobMonitor is ambiguous
> ---------------------------------------------------
>
> Key: HIVE-18672
> URL: https://issues.apache.org/jira/browse/HIVE-18672
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
> Reporter: Sahil Takiar
> Assignee: Sahil Takiar
> Priority: Major
> Attachments: HIVE-18672.1.patch, HIVE-18672.2.patch,
> HIVE-18672.3.patch
>
>
> There are a few places in {{RemoteSparkJobMonitor}} (e.g. when the Spark job
> is in state QUEUED) where the state of the Spark job is printed, but the info
> is ambiguous (no reference to HoS, or the id of the Spark job).
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
