[
https://issues.apache.org/jira/browse/MESOS-2044?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14733687#comment-14733687
]
Kapil Arya commented on MESOS-2044:
-----------------------------------
Exposing IPs as a Slave resource was considered but dropped for the following
reason. The IPs aren't necessarily managed by the Slave. There could be an
external IP address management service available that may not be managed by the
Slave. Further, in some specific cases such as weave (http://weave.works/), an
IP addresses manager instance is available on each Slave node, thus making it
harder for a framework to request it directly.
Further, it seems like there is a lack of clarity in the design doc if it's
giving the impression that either IP-per-container must be supported by all
nodes or not supported at all. This is not the case. The granularity here would
be at the Node level. Either a particular Slave supports ip-per-container, or
it doesn't. In the former case, it would still be possible to launch
tasks/containers without requesting ip-per-container services.
> Use one IP address per container for network isolation
> ------------------------------------------------------
>
> Key: MESOS-2044
> URL: https://issues.apache.org/jira/browse/MESOS-2044
> Project: Mesos
> Issue Type: Epic
> Reporter: Cong Wang
> Assignee: Kapil Arya
> Labels: mesosphere
>
> If there are enough IP addresses, either IPv4 or IPv6, we should use one IP
> address per container, instead of the ugly port range based solution. One
> problem with this is the IP address management, usually it is managed by a
> DHCP server, maybe we need to manage them in mesos master/slave.
> Also, maybe use macvlan instead of veth for better isolation.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
