[
https://issues.apache.org/jira/browse/SPARK-7898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16847651#comment-16847651
]
Kyle Brooks commented on SPARK-7898:
------------------------------------
I have a use case for printing an aggregation does in spark to stdout for use
in a bash script:
{code:java}
from pyspark.sql import SparkSession
spark = SparkSession.builder.getOrCreate()
df = spark.range(10)
count = df.count()
import sys
sys.stderr.write('Why does this get redirected to stdout?\n')
# This is the only thing I want in stdout:
print(count)
{code}
In client mode on a yarn cluster:
{code:java}
spark2-submit --master yarn --deploy-mode client test_print.py 2>stderr.log
1>stdout.log
{code}
When I look in stdout.log:
{code:java}
Why does this get redirected to stdout?
10
{code}
The printing to stderr is done by a third party library I don't have control
over.
Is this behavior by design? I understand that in some modes, the driver will
not run on the machine the job is submitted but for the ones it does, this
seems broken.
> pyspark merges stderr into stdout
> ---------------------------------
>
> Key: SPARK-7898
> URL: https://issues.apache.org/jira/browse/SPARK-7898
> Project: Spark
> Issue Type: Bug
> Components: PySpark
> Affects Versions: 1.3.0
> Reporter: Sam Steingold
> Priority: Major
>
> When I type
> {code}
> hadoop fs -text /foo/bar/baz.bz2 2>err 1>out
> {code}
> I get two non-empty files: {{err}} with
> {code}
> 2015-05-26 15:33:49,786 INFO [main] bzip2.Bzip2Factory
> (Bzip2Factory.java:isNativeBzip2Loaded(70)) - Successfully loaded &
> initialized native-bzip2 library system-native
> 2015-05-26 15:33:49,789 INFO [main] compress.CodecPool
> (CodecPool.java:getDecompressor(179)) - Got brand-new decompressor [.bz2]
> {code}
> and {{out}} with the content of the file (as expected).
> When I call the same command from Python (2.6):
> {code}
> from subprocess import Popen
> with open("out","w") as out:
> with open("err","w") as err:
> p = Popen(['hadoop','fs','-text',"/foo/bar/baz.bz2"],
> stdin=None,stdout=out,stderr=err)
> print p.wait()
> {code}
> I get the exact same (correct) behavior.
> *However*, when I run the same code under *PySpark* (or using
> {{spark-submit}}), I get an *empty* {{err}} file and the {{out}} file starts
> with the log messages above (and then it contains the actual data).
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
