Yarn模式下的Spark文件记录器[英] Spark File Logger in Yarn Mode

问题描述

I want to create a custom logger that writes from messages from executors in a specific folder in a cluster node. I have edited my log4j.properties file in SPARK_HOME/conf/ like this:

log4j.rootLogger=${root.logger}
root.logger=WARN,console
log4j.appender.console=org.apache.log4j.ConsoleAppender
log4j.appender.console.target=System.err
log4j.appender.console.layout=org.apache.log4j.PatternLayout
log4j.appender.console.layout.ConversionPattern=%d{yy/MM/dd HH:mm:ss} %p %c{2}: %m%n
shell.log.level=WARN
log4j.logger.org.eclipse.jetty=WARN
log4j.logger.org.spark-project.jetty=WARN
log4j.logger.org.spark-project.jetty.util.component.AbstractLifeCycle=ERROR
log4j.logger.org.apache.spark.repl.SparkIMain$exprTyper=INFO
log4j.logger.org.apache.spark.repl.SparkILoop$SparkILoopInterpreter=INFO
log4j.logger.org.apache.parquet=ERROR
log4j.logger.parquet=ERROR
log4j.logger.org.apache.hadoop.hive.metastore.RetryingHMSHandler=FATAL
log4j.logger.org.apache.hadoop.hive.ql.exec.FunctionRegistry=ERROR
log4j.logger.org.apache.spark.repl.Main=${shell.log.level}
log4j.logger.org.apache.spark.api.python.PythonGatewayServer=${shell.log.level}

#My logger to write usefull messages in a local file
log4j.logger.jobLogger=INFO, RollingAppenderU

log4j.appender.RollingAppenderU=org.apache.log4j.DailyRollingFileAppender
log4j.appender.RollingAppenderU.File=/var/log/sparkU.log
log4j.appender.RollingAppenderU.DatePattern='.'yyyy-MM-dd
log4j.appender.RollingAppenderU.layout=org.apache.log4j.PatternLayout
log4j.appender.RollingAppenderU.layout.ConversionPattern=[%p] %d %c %M - %m%n
log4j.appender.fileAppender.MaxFileSize=1MB
log4j.appender.fileAppender.MaxBackupIndex=1

I want using the jobLogger to save a file in /var/log/sparkU.log. I created a small program in Python that prints some specific messages:

from pyspark import SparkConf, SparkContext
from pyspark.sql import SQLContext, SparkSession
from pyspark.sql.types import *

spark = SparkSession \
        .builder \
    .master("yarn") \
        .appName("test custom logging") \
        .config("spark.some.config.option", "some-value") \
        .getOrCreate()

log4jLogger = spark.sparkContext._jvm.org.apache.log4j 
log = log4jLogger.LogManager.getLogger("jobLogger") 

log.info("Info message")
log.warn("Warn message")
log.error("Error message")

and I submit it like this:

/usr/bin/spark-submit --master yarn --deploy-mode client /mypath/test_log.py

When I use deploy mode client the file is written at the desired place. When I use deploy mode cluster the local file is not written but the messages can be found in YARN log. But in YARN logs for both modes I take this error also (output for spark cluster mode from YARN logs) :  

log4j:ERROR setFile(null,true) call failed.
java.io.FileNotFoundException: /var/log/sparkU.log (Permission denied)
    at java.io.FileOutputStream.open(Native Method)
    at java.io.FileOutputStream.<init>(FileOutputStream.java:221)
    at java.io.FileOutputStream.<init>(FileOutputStream.java:142)
    at org.apache.log4j.FileAppender.setFile(FileAppender.java:294)
    at org.apache.log4j.FileAppender.activateOptions(FileAppender.java:165)
    at org.apache.log4j.DailyRollingFileAppender.activateOptions(DailyRollingFileAppender.java:223)
    at org.apache.log4j.config.PropertySetter.activate(PropertySetter.java:307)
    at org.apache.log4j.config.PropertySetter.setProperties(PropertySetter.java:172)
    at org.apache.log4j.config.PropertySetter.setProperties(PropertySetter.java:104)
    at org.apache.log4j.PropertyConfigurator.parseAppender(PropertyConfigurator.java:842)
    at org.apache.log4j.PropertyConfigurator.parseCategory(PropertyConfigurator.java:768)
    at org.apache.log4j.PropertyConfigurator.parseCatsAndRenderers(PropertyConfigurator.java:672)
    at org.apache.log4j.PropertyConfigurator.doConfigure(PropertyConfigurator.java:516)
    at org.apache.log4j.PropertyConfigurator.doConfigure(PropertyConfigurator.java:580)
    at org.apache.log4j.helpers.OptionConverter.selectAndConfigure(OptionConverter.java:526)
    at org.apache.log4j.LogManager.<clinit>(LogManager.java:127)
    at org.apache.spark.internal.Logging$class.initializeLogging(Logging.scala:117)
    at org.apache.spark.internal.Logging$class.initializeLogIfNecessary(Logging.scala:102)
    at org.apache.spark.deploy.yarn.ApplicationMaster$.initializeLogIfNecessary(ApplicationMaster.scala:746)
    at org.apache.spark.internal.Logging$class.log(Logging.scala:46)
    at org.apache.spark.deploy.yarn.ApplicationMaster$.log(ApplicationMaster.scala:746)
    at org.apache.spark.deploy.yarn.ApplicationMaster$.main(ApplicationMaster.scala:761)
    at org.apache.spark.deploy.yarn.ApplicationMaster.main(ApplicationMaster.scala)
log4j:ERROR Either File or DatePattern options are not set for appender [RollingAppenderU].
18/01/15 12:13:00 WARN spark.SparkContext: Support for Java 7 is deprecated as of Spark 2.0.0
18/01/15 12:13:02 WARN cluster.YarnSchedulerBackend$YarnSchedulerEndpoint: Attempted to request executors before the AM has registered!
18/01/15 12:13:04 INFO jobLogger: Info message
18/01/15 12:13:04 WARN jobLogger: Warn message
18/01/15 12:13:04 ERROR jobLogger: Error message

So I have two questions:

  -Why the first error message is printed (java.io.FileNotFoundException)? I suspect that this come from the application's master logger but how can I stop it from printing this error? I want only executors to use the file logger.

  -Is it possible to use cluster mode and still be able to write at a specific file in one of the machines? I was wondering If I can somehow enter a path like host:port/myPath/spark.log and all the executors would write in that file in one of the machines.   Thanks in advance for any response.

推荐答案

I was able to use a custom logger to append in a local file in Yarn in cluster mode.

First of all in all cluster worker nodes I made available the log4j file in the same directory (e.g. /home/myUser/log4j.custom.properties ) and also created a folder in the same nodes to save the logs in my user path (e.g. /home/myUser/sparkLogs ).

After that, in submit, I pass that file as the driver logger with driver-java-options and this does the trick. I use this submit (the log4j file is the same as before):

/usr/bin/spark2-submit 
--driver-java-options "-Dlog4j.configuration=file:///home/myUser/log4j.custom.properties"
--master yarn --deploy-mode client --driver-memory nG --executor-memory nG
--executor-cores n /home/myUser/sparkScripts/myCode.py