hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Varun Vasudev (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-4553) Add cgroups support for docker containers
Date Thu, 07 Jan 2016 11:58:40 GMT

    [ https://issues.apache.org/jira/browse/YARN-4553?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15087268#comment-15087268
] 

Varun Vasudev commented on YARN-4553:
-------------------------------------

Patch looks mostly good to me. Minor nits -

1)
{code}
      if (LOG.isInfoEnabled()) {
        LOG.info("no resource restrictions specified. not using docker's "
            + "cgroup options");
      }
    } else {
      if (LOG.isInfoEnabled()) {
        LOG.info("using docker's cgroups options");
      }
{code}
and 
{code}
      if (LOG.isInfoEnabled()) {
        LOG.info("using cgroup parent: " + cGroupPath);
      }
{code}

Maybe these should be log level debug? Seems like a lot of logging for info level.

2)
{code}
    String resourceOptionsCpu = "/sys/fs/cgroup/" + hierarchy + "/cpu/" +
        containerIdStr;
{code}
It shouldn't affect the test but shoudn't it be "/sys/fs/cgroup/cpu/" + hierarchy + "/" +
containerIdStr ?


> Add cgroups support for docker containers
> -----------------------------------------
>
>                 Key: YARN-4553
>                 URL: https://issues.apache.org/jira/browse/YARN-4553
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: yarn
>            Reporter: Sidharta Seethana
>            Assignee: Sidharta Seethana
>         Attachments: YARN-4553.001.patch, YARN-4553.002.patch
>
>
> Currently, cgroups-based resource isolation does not work with docker containers under
YARN. The processes in these containers are launched by the docker daemon and they are not
children of a container-executor process. Docker supports a --cgroup-parent flag which can
be used to point to the container-specific cgroups that are created by the nodemanager. This
will allow the Nodemanager to manage cgroups (as it does today) while allowing resource isolation
to work with docker containers. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message