You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mesos.apache.org by "Gilbert Song (JIRA)" <ji...@apache.org> on 2016/04/06 07:45:25 UTC

[jira] [Commented] (MESOS-5127) Reset `LIBPROCESS_IP` in `network\cni` isolator.

    [ https://issues.apache.org/jira/browse/MESOS-5127?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15227765#comment-15227765 ] 

Gilbert Song commented on MESOS-5127:
-------------------------------------

Do we have the same issue in custom executor? Since we have a new network namespace created by CNI isolator, it seems to me that setting the `LIBPROCESS_IP` as 0.0.0.0 should also fix the same case in custom executor.

> Reset `LIBPROCESS_IP` in `network\cni` isolator.
> ------------------------------------------------
>
>                 Key: MESOS-5127
>                 URL: https://issues.apache.org/jira/browse/MESOS-5127
>             Project: Mesos
>          Issue Type: Bug
>          Components: containerization
>         Environment: Linux
>            Reporter: Avinash Sridharan
>            Assignee: Avinash Sridharan
>
> Currently the `LIBPROCESS_IP` environment variable was being set to
>     the Agent IP if the environment variable has not be defined by the
>     `Framework`. For containers having their own IP address (as with
>     containers on CNI networks) this becomes a problem since the command
>     executor tries to bind to the `LIBPROCESS_IP` that does not exist in
>     its network namespace, and fails. Thus, for containers launched on CNI
>     networks the `LIBPROCESS_IP` should not be set, or rather is set to
>     "0.0.0.0", allowing the container to bind to the IP address provided
>     by the CNI network.



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