[
https://issues.apache.org/jira/browse/MESOS-6202?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15504216#comment-15504216
]
Anand Mazumdar commented on MESOS-6202:
---------------------------------------
We undertook the work to clean up orphaned docker containers correctly as part
of MESOS-3573. Unfortunately, this modified behavior should have been part of
the 1.0 {{CHANGELOG}} but somehow was missed. As [[email protected]]
suggested, you can use the {{docker_kill_orphans}} to get the previous
behavior.
We currently don't look to see if the {{id}} also has a valid UUID as you had
pointed out. It seems orthogonal to this issue though. [~h0tbird] Can you file
a separate issue for that?
> Docker containerizer kills containers whose name starts with 'mesos-'
> ---------------------------------------------------------------------
>
> Key: MESOS-6202
> URL: https://issues.apache.org/jira/browse/MESOS-6202
> Project: Mesos
> Issue Type: Bug
> Components: containerization, docker
> Affects Versions: 1.0.1
> Environment: Dockerized
> {{mesosphere/mesos-slave:1.0.1-2.0.93.ubuntu1404}}
> Reporter: Marc Villacorta
>
> I run 3 docker containers in my CoreOS system whose names start with
> _'mesos-'_ those are: _'mesos-master'_, _'mesos-dns'_ and _'mesos-agent'_.
> I can start the first two without any problem but when I start the third one
> _('mesos-agent')_ all three containers are killed by the docker daemon.
> If I rename the containers to _'m3s0s-master'_, _'m3s0s-dns'_ and
> _'m3s0s-agent'_ everything works.
> I tracked down the problem to
> [this|https://github.com/apache/mesos/blob/16a563aca1f226b021b8f8815c4d115a3212f02b/src/slave/containerizer/docker.cpp#L116-L120]
> code which is marked to be removed after deprecation cycle.
> I was previously running Mesos 0.28.2 without this problem.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
