[
https://issues.apache.org/jira/browse/SPARK-18829?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15744464#comment-15744464
]
David Hodeffi commented on SPARK-18829:
---------------------------------------
The issue is it's hard to debug application on yarn. stdout file is **not** the
same file as the logger file.
option1 : update the function to get arg of type OutputStream and then write
string to it but its overkill.
option2: simply, go to log4j.properties and configure the logger to print that
out.
right now option1 / option2 is not possible, that's why I suggested to print it
to logger too.
changing standard output just because you want to see just this specific log is
not is not a good idea, because it will spam the log file with everything that
is printed to standard output and you cannot really control that.
> Printing to logger
> ------------------
>
> Key: SPARK-18829
> URL: https://issues.apache.org/jira/browse/SPARK-18829
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core
> Affects Versions: 1.6.2
> Environment: ALL
> Reporter: David Hodeffi
> Priority: Trivial
> Labels: easyfix, patch
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> I would like to print dataframe.show or df.explain(true) into log file.
> right now the code print to standard output without a way to redirect it.
> It also cannot be configured on log4j.properties.
> My suggestion is to write to the logger and standard output.
> i.e
> class DataFrame {......
> override def explain(extended: Boolean): Unit = {
> val explain = ExplainCommand(queryExecution.logical, extended = extended)
> sqlContext.executePlan(explain).executedPlan.executeCollect().foreach {
> // scalastyle:off println
> r => {
> println(r.getString(0))
> logger.debug(r.getString(0))
> }
> }
> // scalastyle:on println
> }
> }
> def show(numRows: Int, truncate: Boolean): Unit = {
> val str =showString(numRows, truncate)
> println(str)
> logger.debug(str)
> }
> }
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
