You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Chao Sun (JIRA)" <ji...@apache.org> on 2017/06/28 21:05:00 UTC

[jira] [Comment Edited] (HIVE-16962) Better error msg for Hive on Spark in case user cancels query and closes session

    [ https://issues.apache.org/jira/browse/HIVE-16962?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16067221#comment-16067221 ] 

Chao Sun edited comment on HIVE-16962 at 6/28/17 9:04 PM:
----------------------------------------------------------

Looking good. Just a few nits:
1. maybe consistently use "session" or "Session" in the log messages.
2. line #70 of SparkSessionImpl: extra space after "opened".
3. line #140 of SparkSessionImpl: control flow could reach here after catching the IOException, and so there will be two log messages: "Failed to close spark session ..." and "Spark session .. is successfully closed". This is confusing.
4. line #130 of SparkClientImpl: maybe add the error 'ie' after the error message, so we can get the stacktrace. Also, in case of user cancellation, I'm not sure if we need to log it at ERROR level.


was (Author: csun):
Looking good. Just a few nits:
1. maybe consistently use "session" or "Session" in the log messages.
2. line #70 of SparkSessionImpl: extra space after "opened".
3. line #140 of SparkSessionImpl: control flow could reach here after catching the IOException, and so there will be two log messages: "Failed to close spark session ..." and "Spark session .. is successfully closed". This is confusing.
3. line #130 of SparkClientImpl: maybe add the error 'ie' after the error message, so we can get the stacktrace. Also, in case of user cancellation, I'm not sure if we need to log it at ERROR level.

> Better error msg for Hive on Spark in case user cancels query and closes session
> --------------------------------------------------------------------------------
>
>                 Key: HIVE-16962
>                 URL: https://issues.apache.org/jira/browse/HIVE-16962
>             Project: Hive
>          Issue Type: Improvement
>          Components: Spark
>    Affects Versions: 1.1.0
>            Reporter: Xuefu Zhang
>            Assignee: Xuefu Zhang
>         Attachments: HIVE-16962.patch, HIVE-16962.patch
>
>
> In case user cancels a query and closes the session, Hive marks the query as failed. However, the error message is a little confusing. It still says:
> {quote}
> org.apache.hive.service.cli.HiveSQLException: Error while processing statement: FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.spark.SparkTask. Failed to create spark client. This is likely because the queue you assigned to does not have free resource at the moment to start the job. Please check your queue usage and try the query again later.
> {quote}
> followed by some InterruptedException.
> Ideally, the error should clearly indicates the fact that user cancels the execution.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)