-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/39769/#review105516
-----------------------------------------------------------



src/slave/slave.cpp (line 3268)
<https://reviews.apache.org/r/39769/#comment164139>

    Right here, we need to reset the rootfs image. I will just add the 
following comment+code and merge:
    
    ```
         // For command-tasks, we are now copying the entire `task.container` 
into
         // the `executorInfo`. Thus, `executor.container` now has the image if
         // `task.container` had one. However, in case of rootfs, we want to run
         // the command executor in the host filesystem and prepare/mount the 
image
         // into the container as a volume (command executor will use 
pivot_root to
         // mount the image). For this reason, we need to strip the image in
         // `executor.container.mesos`.
         container->mutable_mesos()->clear_image();
    ```


- Kapil Arya


On Nov. 6, 2015, 4:58 p.m., Spike Curtis wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/39769/
> -----------------------------------------------------------
> 
> (Updated Nov. 6, 2015, 4:58 p.m.)
> 
> 
> Review request for mesos, Jie Yu and Kapil Arya.
> 
> 
> Bugs: MESOS-3810
>     https://issues.apache.org/jira/browse/MESOS-3810
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Always copy ContainerInfo to ExecutorInfo
> 
> Remove MesosContainerizer check for TaskInfo->ContainerInfo
> 
> 
> Diffs
> -----
> 
>   src/slave/slave.cpp 7d70e86e03d525fc16827cc36ed55c250ca2735b 
>   src/tests/slave_tests.cpp 6b89eaf0e5c346183dbff90adb4a560b9342ca6f 
> 
> Diff: https://reviews.apache.org/r/39769/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> passes network testing framework battery of tests at 
> https://github.com/mesosphere/net-modules/tree/integration/0.25
> 
> 
> Thanks,
> 
> Spike Curtis
> 
>

Reply via email to