You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Aleksandr Balitsky (JIRA)" <ji...@apache.org> on 2016/09/06 14:01:20 UTC
[jira] [Updated] (YARN-5619) Provide way to limit MRJob's
stdout/stderr size
[ https://issues.apache.org/jira/browse/YARN-5619?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Aleksandr Balitsky updated YARN-5619:
-------------------------------------
Description:
We can run job with huge amount of stdout/stderr and causing undesired consequence. There is already a Jira which is been open for while now:
https://issues.apache.org/jira/browse/YARN-2231
The possible solution is to redirect Stdout's and Stderr's output to log4j in YarnChild.java main method via commands:
System.setErr( new PrintStream( new LoggingOutputStream( <Log4j logger>, Level.ERROR ), true));
System.setOut( new PrintStream( new LoggingOutputStream( <Log4j logger>, Level.INFO ), true));
In this case System.out and System.err will be redirected to log4j logger with appropriate appender that will direct output to stderr or stdout files with needed size limitation.
Advantages of such solution:
- it allows us to restrict file sizes during job execution.
Disadvantages:
- It will work only for MRs jobs.
- logs are stored in memory and are flushed on disk only after job's finishing (syslog works the same way) - we can loose logs if container will be killed or failed.
Is it appropriate solution for solving this problem, or is there something better?
was:
We can run job with huge amount of stdout/stderr and causing undesired consequence. There is already a Jira which is been open for while now:
https://issues.apache.org/jira/browse/YARN-2231
The possible solution is to redirect Stdout's and Stderr's output to log4j in YarnChild.java main method via commands:
System.setErr( new PrintStream( new LoggingOutputStream( <Log4j logger>, Level.ERROR ), true));
System.setOut( new PrintStream( new LoggingOutputStream( <Log4j logger>, Level.INFO ), true));
In this case System.out and System.err will be redirected to log4j logger with appropriate appender that will direct output to stderr or stdout files with needed size limitation.
Advantages of such solution:
- it allows us to restrict file sizes during job execution.
Disadvantages:
- It will work only for MRs jobs.
- logs are stored in memory and are flushed on disk only after job's finishing (syslog works the same way) - we can loose logs if container will be killed or failed.
Is it appropriate solution for solving this problem, or is there something better?
> Provide way to limit MRJob's stdout/stderr size
> -----------------------------------------------
>
> Key: YARN-5619
> URL: https://issues.apache.org/jira/browse/YARN-5619
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: log-aggregation, nodemanager
> Affects Versions: 2.7.0
> Reporter: Aleksandr Balitsky
> Priority: Minor
>
> We can run job with huge amount of stdout/stderr and causing undesired consequence. There is already a Jira which is been open for while now:
> https://issues.apache.org/jira/browse/YARN-2231
> The possible solution is to redirect Stdout's and Stderr's output to log4j in YarnChild.java main method via commands:
> System.setErr( new PrintStream( new LoggingOutputStream( <Log4j logger>, Level.ERROR ), true));
> System.setOut( new PrintStream( new LoggingOutputStream( <Log4j logger>, Level.INFO ), true));
> In this case System.out and System.err will be redirected to log4j logger with appropriate appender that will direct output to stderr or stdout files with needed size limitation.
> Advantages of such solution:
> - it allows us to restrict file sizes during job execution.
> Disadvantages:
> - It will work only for MRs jobs.
> - logs are stored in memory and are flushed on disk only after job's finishing (syslog works the same way) - we can loose logs if container will be killed or failed.
> Is it appropriate solution for solving this problem, or is there something better?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org