A set of scripts we use with Docker and Amazon EC2 Container Service (ECS below)/ECR to manage things.
In general, our continuous integration process goes like this:
- Check out new version
- Build everything
- Run unit tests
- Run protractor tests (automated in-browser integration tests for angularjs) (non deployment branches stop here)
- Build a docker image
- Push that image to ECR
- Use update-tasks
- Use update-service
- Run protractor tests against live server
This is part of our continuous integration environment to deploy changes. We use this script in our build process after we build, tag, and push our docker images.
This script will register a new version of an ECS task definition so it can be used to update a service in our docker cluster.
It copies all settings from the most recent task definition, and sets a new image.
To make it work, we tag our images with a version number in the docker registry in the format of v## (ex. v10 or v100) (We grab the build number form continuous integration environment variable representing the build number.)
You should set AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY env variables. Optional to set AWS_EC2_REGION (defaults to us-east-1)
usage: update-tasks.py [-h] [--tag TAG] [--tasks TASKS [TASKS ...]]
[--images IMAGES [IMAGES ...]]
Register new versions of container images in tasks.
optional arguments:
-h, --help show this help message and exit
--tag TAG Docker tag of new version of images
--tasks TASKS [TASKS ...] ECS Task Definitions to update
--images IMAGES [IMAGES ...] Container images to update within the tasks
Example:
python update-tasks.py --task web celery --images scrumdo-web --tag v248
This would upgrade two tasks, web and celery. Inside those tasks, it would set any container using the scrumdo-web image to use the one tagged v248
NOTE: This script does not update any services, so you have to either push a button or run another script to actually do the deployment.
This is part of our continuous integration environment to deploy changes.
This script will take an ECS service, look up any updated task definitions for the task, and update the service with the new version of the task definitions which will cause ECS to go through it's deploy process.
After performing the update, this script will wait up to 10 minutes for the deployment to finish. We use this to wait to announce the deployment and to start a round of automated tests on the live servers.
This script is only safe to run if you know that your latest task definition is the one you will want. Don't use it if you use the same taskdef for multiple environments.
usage: update-service.py [-h] --cluster CLUSTER --service SERVICE
Register new versions of task definitions in a service.
optional arguments:
-h, --help show this help message and exit
--cluster CLUSTER ECS cluster to work on
--service SERVICE Service to update
We run this script from inside a container in a startup script. It sets a DNS lookup to itself, utilizing the EC2 meta-info to figure out it's own ip address. This is a cheap and easy way to do service discovery, there are better ways since changing the host that the service runs on could result in downtime equal to your TTL. (currently, we use this for a cheap development/qa redis server)
usage: route53-presence.py [-h] [--ttl TTL] [--local] hostname
Register or unregister a name in Route53.
positional arguments:
hostname fqdn to manipulate
optional arguments:
-h, --help show this help message and exit
--ttl TTL ttl in seconds, default 600
--local use local IP instead of public
Example:
./route53-presence.py --ttl 60 --local myservice.domain.com
Registers our local/private (10.*.*.*) address with route53 to the given hostname with a time to live of 60 seconds.
No copyright claimed on this script, original version of this script taken from: https://github.com/timelinelabs/docker-route53-presence