Deploy stateful applications with Docker Swarm

You can use Portworx to provide storage for your Docker Swarm services. Portworx pools your servers capacity and turns your servers or cloud instances into converged, highly available compute and storage nodes. This section describes how to deploy Portworx within a Docker Swarm cluster and have Portworx provide highly available volumes to any application deployed via Docker Swarm.

Install Portworx

Below steps demonstrate how to set up a three-node cluster for Jenkins and use a Portworx volume.

Create a volume

docker volume create -d pxd --name jenkins_vol --opt \
        size=4 --opt block_size=64 --opt repl=3 --opt fs=ext4 --opt sharedv4=true

• This command creates a volume called jenkins_vol.
• This volume has a replication factor of 3, which means that the data will be protected on 3 separate nodes.
• Also the volume is shared so multiple swarm nodes can have shared access

You can inspect the jenkins_vol volume using the pxctl CLI:

pxctl volume inspect jenkins_vol

    Volume : 27052673284397061
    Name : jenkins_vol
    Size : 4.0 GiB
    Format : ext4
    HA : 3
    IO Priority : LOW
    Creation time : Apr 4 22:23:32 UTC 2017
    Shared : yes
    Status : up
    State : detached
    Reads : 0
    Reads MS : 0
    Bytes Read : 0
    Writes : 0
    Writes MS : 0
    Bytes Written : 0
    IOs in progress : 0
    Bytes used : 130 MiB
    Replica sets on nodes:
        Set 0
            Node : 192.168.56.103
            Node : 192.168.56.104
            Node : 192.168.56.105

(Optional) Create node labels for convergence

• Identify the nodes where the replica set of the jenkins_vol volume resides using output of inspect command above.

• Add a label to each of these nodes as below. This will later allow us to create a service whose tasks only run on these nodes.

  docker node update --label-add jenkins_vol=true <node_name>

Create a service

We will now create a Jenkins service using the newly created volume.

We will use service constraints to influence on which worker node Swarm schedules a container (task) based on the container volume’s data location.

docker service create --name jenkins \
         --replicas 3 \
         --publish 8082:8080 \
         --publish 50000:50000 \
         -e JENKINS_OPTS="--prefix=/jenkins" \
         --reserve-memory 300m \
         --mount "type=volume,volume-driver=pxd,source=jenkins_vol,target=/var/jenkins_home" \
         --constraint 'node.labels.jenkins_vol == true' \
         jenkins

• Note how the volume binding is done via --mount. This causes the Portworx jenkins_vol to get bind mounted at /var/jenkins_home, which is where the jenkins Docker container stores it’s data.
• Make sure you specify volume-driver=pxd in the --mount option. This ensures that docker always uses jenkins_vol provided the Portworx’s pxd volume driver
• Also note how we put a constraint using --constraint 'node.labels.jenkins_vol == true'.

Now Docker Swarm will place the jenkins container only on Swarm nodes that contain our volume’s data locally leading to great I/O performance.

Verify Service

Use following command to verify if various tasks for the service came up:

docker service ps jenkins

Read more about the Portworx Docker Swarm demo here.