Setting up a Timemachine server using Mesos/Marathon

I am currently unable to upgrade to the latest version of Slack due to this bug. That combined with my user profile confusion (originally I set up using a US account, then changed to an South African one) that causes requests to update to versions of iPhoto which aren’t available in my region made me decide that it is time for a reformat.

The Problem

To create a full backup to my raid’ed server, I need to set up netatalk. I started reading this post but decided that I didn’t want to pollute my server with all of those installs. Seeing that I already have Mesos and Marathon running, I went the docker route. I found this repo and proceeded to create the following Marathon config:

{
  "id": "netatalk",
  "instances": 1,
  "cpus": 0.2,
  "mem": 256.0,
  "constraints": [["hostname", "CLUSTER", "myhost.mydoamin.com"]],
  "container": {
    "docker": {
      "image": "cptactionhank/netatalk:latest",
      "network": "HOST",
      "portMappings":
      [
          {
              "containerPort": 548,
              "protocol": "tcp"
          }
      ]
    },
    "type": "DOCKER",
    "volumes": [
      {
        "hostPath": "/data/media",
        "containerPath": "/media/share",
        "mode": "RW"
      },
      {
        "hostPath": "/data/timemachine",
        "containerPath": "/media/timemachine",
        "mode": "RW"
      }
    ]
  },
  "env": {
    "AVAHI": "1",
    "AFP_USER": "timemachine",
    "AFP_PASSWORD": "super_secret",
    "AFP_UID": "1012",
    "AFP_GID": "1007"
  },
  "healthChecks": [
    {
      "gracePeriodSeconds": 60,
      "intervalSeconds": 30,
      "maxConsecutiveFailures": 0,
      "port": 548,
      "protocol": "TCP"
    }
  ]
}

The important parts

Limiting where the container may run, in my case, I only want it on my server with the raid array:

"constraints": [["hostname", "CLUSTER", "myhost.mydoamin.com"]]

Netatalk runs on port tcp/548:

"containerPort": 548

If you want to use service discovery, you need to have the container run on the host network stack:

"network": "HOST"

You can also use netatalk to share media easily, so I mounted my media folder to share as well:

"hostPath": "/data/media",
"containerPath": "/media/share",
"mode": "RW"

The folder that timemachine will use to back up:

"hostPath": "/data/timemachine",
"containerPath": "/media/timemachine",
"mode": "RW"

The environment variables for the container:

"env": {
  "AVAHI": "1",
  "AFP_USER": "timemachine",
  "AFP_PASSWORD": "super_secret",
  "AFP_UID": "1012",
  "AFP_GID": "1007"
},

Service discovery is set with AVAHI=1 and the other values are for credentials for timemachine. You will need to create a user on the server and get the userid and groupid for it, i.e.:

sudo adduser --home /mnt/data/timemachine timemachine
sudo chown -R timemachine:timemachine /mnt/data/timemachine

To get the userid and groupid to use, run the following:

sudo -H -u timemachine bash -c 'echo "I am $(id -un), with uid $(id -u) and gid $(id -g)"'

And that is all folks, it takes 5 minutes to set up a timemachine server using Docker, Mesos and Marathon. After the container has started up, head over to timemachine and you should see it listed when you click on the Select disk... button. Now I just need to add the user creation to my Chef recipes so I don’t forget about it when I recreate the node…