Triangle }}

Docker on Mesos

With the help of our Docker executor for Mesos, Mesos can run and manage Docker containers in conjunction with Chronos and Mesosphere’s Marathon frameworks. Our executor is the connector between Mesos and the underlying Docker task.

Docker containers provide a consistent, compact and flexible means of packaging application builds. Delivering applications with Docker on Mesos promises a truly elastic, efficient and consistent platform for delivering a range of applications on premises or in the cloud.

Service Launch and Discovery

Let’s launch a service in a Docker container and use Marathon’s discovery features to find and connect to it from another Docker container. In this example, we’ll launch Redis within a Docker container and then connect to it using the Redis CLI.

Launching Docker containers with Marathon is a matter of assigning the “executor” parameter to point to a Docker executor and making the container name the first word of the command.

:;  http POST http://localhost:8080/v1/apps/start 
              id=multidis instances=2 mem=512 cpus=1 
              executor=/var/lib/mesos/executors/docker 
              cmd='johncosta/redis'

>>

HTTP/1.1 204 No Content
Content-Type: application/json
Server: Jetty(8.y.z-SNAPSHOT)

Once we have two Redis instances running, we can query Marathon to find them and connect to them.

:;  http GET http://localhost:8080/v1/endpoints

>>

HTTP/1.1 200 OK
Content-Type: text/plain
Server: Jetty(8.y.z-SNAPSHOT)
Transfer-Encoding: chunked

multidis 10445 ip-10-184-7-73.ec2.internal:31001 ip-10-184-7-73.ec2.internal:31000

Using one of the ip:port pairs above, we can connect to one of our Redis instances, using the same Docker image, to get access to a working Redis CLI:

:;  sudo docker run -i -t johncosta/redis 
         redis-cli -h ip-10-184-7-73.ec2.internal -p 31001

redis ip-10-184-7-73.ec2.internal:31001> SET x 7
OK
redis ip-10-184-7-73.ec2.internal:31001> GET x
"7"

There’s no need to install Redis or its supporting libraries on your Mesos hosts.

Mesos And Docker In Operation

A Mesos cluster comprises a few masters and many slaves. As work is farmed out to them, Mesos slaves delegate to executors for the setup and teardown of individual tasks. It is the executor that manages communication between the slave and the Docker daemon. The Docker daemon manages caching and launch of Docker images, which can be pulled from the global Docker index or a local registry.

Marathon can launch and monitor service containers from one or more
      Docker registries using the Docker executor for Mesos.

When a Docker container is started as a Mesos task, it runs beneath the Docker daemon on a slave server. Although the Docker container does not run as a true child process of the executor, as it would have under the old Docker standalone mode, the executor is able to manage translation of resource limits, signals and Mesos process control messages to appropriate calls of the docker tool. Fine-grained resource monitoring, forthcoming in mainline Mesos and part of some service management tools, is hard to do right if the container does not run directly under the executor — so this architecture may be revised in the future.

The executor, monitored by the Mesos slave, delegates to the local
      Docker daemon for image discovery and management. The executor
      communicates with Marathon via the Mesos master and ensures that Docker
      enfores the specified resource limitations.

When a user requests a Docker container with Marathon, using the web UI or the HTTP API as in the examples above, the request is ultimately delegated to the LXC tools, by way of Mesos and Docker.

  • Marathon makes a resource request from the Mesos master and then waits to accept an appropriate offer.

  • Once an offer is accepted, the Mesos master sends the task’s specification to the slave.

  • On the slave server, the Mesos slave daemon calls the Mesos-Docker executor which in turn calls the docker command line tool.

  • The docker command line tool talks to the local Docker daemon, which manages interactions with the cache of images and the LXC tools.

  • If the image is cached, it will be run from the cache. Otherwise, the Docker daemon contacts a Docker registry to retrieve it.

  • The Docker daemon runs the container under the LXC tools.

Reproducing This Example

Installation of the many components used here is straightforward. Mesos and the Python bindings for Mesos are available as packages:

Marathon, a Scala program, can be run from a standalone Jar. We’ve made an Upstart script for it, too.

With Mesos and Marathon, you have everything you need to start running reliable services across your cluster.

The Docker project provides Ubuntu 13.04 installation instructions.

To follow along with the examples above, install the Docker executor as /var/lib/mesos/executors/docker.

Docker Executor for Mesos

For a standalone installation, we’ve put together a script that installs all the needed components and configures them on Ubuntu 13.04. Just pipe it to a shell running under sudo and you’re all set:

curl -fL https://raw.github.com/mesosphere/mesos-docker/master/bin/mesos-docker-setup | sudo bash