Skip to content

Latest commit

 

History

History
125 lines (99 loc) · 6.78 KB

docker.md

File metadata and controls

125 lines (99 loc) · 6.78 KB
Raw

Docker and Infrastructure Dependencies

In KillrVideo, we offer microservice implementations in multiple programming languages, but each of these implementations have a base common set of dependencies they need in order to be a functional reference application. Those base dependencies are:

  • DataStax Enterprise: we need at least one node of DSE / Cassandra running that the microservices implementations can use to store data.
  • KillrVideo Web Tier: we need the Web Tier in order to provide a working UI and application that drives calls to the microservices.
  • etcd: we use etcd as our service registry for service discovery so that our microservices know where to find DSE and our Web Tier knows where to find our microservices.

For a developer looking to learn about Cassandra and DSE it would be impractical (and kind of ridiculous) to ask them to install all of these things locally in addition to the code for the microservices implementation they are interested in. Instead, we've chosen to use Docker to run these dependencies. Docker has become an increasingly popular way for developers to run infrastructure on their local dev machines. This way, the only thing a developer needs to install is Docker for Mac or Windows in order to try out KillrVideo (and chances are they may already have it installed).

Docker Compose for Starting Dependencies

The Docker project offers a nice tool called Docker Compose which allows you to run multiple Docker containers that are defined in a .yaml file. We created the killrvideo-docker-common project to house the configuration that is the same across many of our projects. This project includes the following:

  • A base docker-compose.yaml configuration that includes a single DSE node using the official DSE image, which is available in the Docker Store. It also includes etcd and Registrator, a program that automatically registers and deregisters Docker containers with the etcd service registry when they are started and stopped.
  • Additional docker-compose files that show how to use Docker images for DataStax tools including Studio and OpsCenter available in Docker Hub.
  • Helper scripts (for Windows and Mac) that can be used to determine the Docker network setup on a developer's machine so we know what IP addresses the components can use to communicate with each other.

You might notice that the Web Tier is not listed as one of the dependencies specified in the base docker-compose.yaml file. This is because the killrvideo-web project also depends on the killrvideo-docker-common project to spin up dependencies when it's being developed. In the microservice implementations, we take advantage of a feature of Docker Compose that lets us define a .yaml file that inherits from our base .yaml file and then provide overrides and additions. In this way, each programming language's microservice project can specify the Web Tier as a dependency (along with any other infrastructure they might need, for example, messaging infrastructure, or DataStax Studio.

Docker Environment Scripts

As mentioned above, the killrvideo-docker-common project also contains some scripts that are useful for determining a developer's Docker setup. These scripts can be used to create a .env file which is used by Docker Compose when it runs to set environment variables needed by KillrVideo. That .env file can also be used by the various microservice implementations to make sure that they register themselves with the appropriate IP address in etcd for service discovery. The contents of the .env file include:

  • COMPOSE_PROJECT_NAME: the name that will prefix each container name (i.e. killrvideo_dse_1).
  • COMPOSE_FILE: the list of compose files to use, for example: a compose file from killrvideo-docker-common as well as a local compose file.
  • KILLRVIDEO_DOCKER_IP: the Docker VM's IP where the containers are hosted. This is where the Web UI, DSE node, and etcd are available.
  • KILLRVIDEO_HOST_IP: the IP address of the user's local dev machine. This is the IP address that the microservices should use to listen on and to register themselves with etcd.
  • KILLRVIDEO_DOCKER_TOOLBOX: set to either true or false depending on whether the local docker environment is a Docker Toolbox setup.

Once we have the docker-compose.yaml files described above, as well as a .env file from running the helper scripts, it's easy to start all our infrastructure dependencies by simply running:

> docker-compose up

Other Docker Compose Examples

You may be interested in other configuration options for running KillrVideo in Docker, such as:

  • Starting a copy of OpsCenter, the web-based visual management and monitoring tool, to monitor your KillrVideo DSE node(s).
  • Starting a copy of Studio, an interactive web-based developer tool for interacting with Cassandra data using CQL and Graph data using Gremlin.
  • Storing DSE / Cassandra data on a volume mounted from the host machine, so that your data can survive beyond the lifespan of a single container.

These alternate configurations are available as part of the killrvideo-docker-common repository, which is included as part of the service implemementation repositories.

To use one of these alternate configurations, you'll want to edit the COMPOSE_FILE variable in the .env file described above to include the docker-compose of your choice instead of the basic docker-compose.yaml file.

For example, you might change the default configuration of:

COMPOSE_FILE=./lib/killrvideo-docker-common/docker-compose.yaml:./docker-compose.yaml

to:

COMPOSE_FILE=./lib/killrvideo-docker-common/docker-compose-volumes.yaml:./docker-compose.yaml

in order to use the docker-compose-volumes.yaml file which starts a DSE instance with the data drive mounted from the host machine.

Next, we'll take a quick look at service discovery in KillrVideo with etcd.

Next: Service Discovery with etcd