Dockercoins On Swarm

By the end of this exercise, you should be able to:

• Deploy an application on swarm as a 'stack', using a docker compose file
• Get some high-level monitoring information about the services and tasks running as part of a stack

Deploying a Stack

1. Deploy Dockercoins as a stack on your swarm, from node-0:

   [centos@node-0 ~]$ cd ~/orchestration-workshop/dockercoins
   [centos@node-0 dockercoins]$ docker stack deploy -c=docker-compose.yml dc
   

2. Check and see how your services are doing:

   [centos@node-0 dockercoins]$ docker stack services dc
   
   ID            NAME       MODE        REPLICAS  IMAGE                             PORTS
   7tcaa3d3g9d2  dc_webui   replicated  1/1       training/dockercoins_webui:1.0    *:8000->80/tcp
   mfq8i9cr4rcj  dc_rng     replicated  1/1       training/dockercoins_rng:1.0      *:8001->80/tcp
   sbywya5yyrus  dc_redis   replicated  1/1       redis:latest                       
   tk1ydqu794ng  dc_hasher  replicated  1/1       training/dockercoins_hasher:1.0   *:8002->80/tcp
   v6835nwsek2n  dc_worker  replicated  1/1       training/dockercoins_worker:1.0
   

   Notice the REPLICAS column in the output of above command; this shows how many of your desired replicas are running. At first, a few might show 0/1; before those tasks can start, the worker nodes will need to download the appropriate images from Docker Hub.

3. Wait a minute or two, and try docker stack services dc again; once all services show 100% of their replicas are up, things are running properly and you can point your browser to port 8000 on one of the swarm nodes (does it matter which one?). You should see a graph of your dockercoin mining speed, around 3 hashes per second.

4. Finally, check out the details of the tasks running in your stack with stack ps:

   [centos@node-0 dockercoins]$ docker stack ps dc
   

   This shows the details of each running task scheduled by services in your stack, similar to service ps, but for each service in the stack. Notice that these containers have been scheduled across our swarm, not just on one node like Docker Compose did.

Conclusion

In this exercise, we stood up our first stack. A stack is a collection of docker components (services, networks, volumes, etc) that make up a full application, and we can create one directly from the same docker compose file we used to start an application on a single host using docker compose. A stack, however, will schedule workloads using Swarm, distributed across our cluster.