[
https://issues.apache.org/jira/browse/MESOS-5727?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15434727#comment-15434727
]
Alexander Rukletsov commented on MESOS-5727:
--------------------------------------------
Though the benefits of sharing the mount namespace between the command executor
and its task are clear, there are disadvantages as well.
Once we {{pivot_root}}, we may loose access to helper and system binaries,
which we may need in the executor for, e.g., health checking. Moreover, in the
future pods case, an executor will have multiple tasks with their own mount
namespaces, hence the executor must have its own mount namespace.
We will be {{unshare}}'ing the task's mount namespace:
https://reviews.apache.org/r/51266/
> Command executor health check does not work when the task specifies container
> image.
> ------------------------------------------------------------------------------------
>
> Key: MESOS-5727
> URL: https://issues.apache.org/jira/browse/MESOS-5727
> Project: Mesos
> Issue Type: Bug
> Components: containerization
> Affects Versions: 0.28.2, 1.0.0
> Reporter: Jie Yu
> Assignee: Gilbert Song
> Labels: containerizer, health-check, mesosphere
> Fix For: 1.0.0
>
>
> Since we launch the task after pivot_root, we no longer has the access to the
> mesos-health-check binary. The solution is to refactor health check to be a
> library (libprocess) so that it does not depend on the underlying filesystem.
> One note here is that we should strive to keep both the command executor and
> the task in the same mount namespace so that Mesos CLI tooling does not need
> to find the mount namespace for the task. It just need to find the
> corresponding pid for the executor. This statement is *arguable*, see the
> comment below.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
