hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eric Badger (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (YARN-6091) the AppMaster register failed when use Docker on LinuxContainer
Date Fri, 07 Apr 2017 15:42:41 GMT

     [ https://issues.apache.org/jira/browse/YARN-6091?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Eric Badger updated YARN-6091:
    Attachment: YARN-6091.001.patch

Uploading a patch that redirects stdout to /dev/null for the popen commands whose stdout we
don't read. I tested this change locally leveraging the test-container-executor program. When
you don't redirect stdout you get a return 13, which means SIGPIPE. When you do redirect,
you get a return value of 0. 

However, test-container-executor doesn't run without error, so I had to work around errors
and comment out large pieces of the code to test the relevant section of in launch_docker_container_as_user
that uses popen()/pclose(). We might need a new JIRA to fix test-container-executor. I'm not
sure what the plan on it is going forward, since it's not a part of the maven testing. At
this point, it's clearly out of sync so we need to either scrap it or maintain it. 

For context, I tried to run test-container-executor on both macOS and rhel7 and they both
failed in separate ways. 

> the AppMaster register failed when use Docker on LinuxContainer 
> ----------------------------------------------------------------
>                 Key: YARN-6091
>                 URL: https://issues.apache.org/jira/browse/YARN-6091
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: nodemanager, yarn
>    Affects Versions: 2.8.1
>         Environment: CentOS
>            Reporter: zhengchenyu
>            Priority: Critical
>         Attachments: YARN-6091.001.patch
>   Original Estimate: 336h
>  Remaining Estimate: 336h
> In some servers, When I use Docker on LinuxContainer, I found the aciton that AppMaster
register to Resourcemanager failed. But didn't happen in other servers. 
> I found the pclose (in container-executor.c) return different value in different server,
even though the process which is launched by popen is running normally. Some server return
0, and others return 13. 
> Because yarn regard the application as failed application when pclose return nonzero,
and yarn will remove the AMRMToken, then the AppMaster register failed because Resourcemanager
have removed this applicaiton's token. 
> In container-executor.c, the judgement condition is whether the return code is zero.
But man the pclose, the document tells that "pclose return -1" represent wrong. So I change
the judgement condition, then slove this problem. 

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org

View raw message