This tool aims to simplify the deployment of docker-based services on Amazon's ECS platform.
Creation/deployment of ECS services is backed by AWS Cloud Formation.
Every command in the tool maps to a stack operation in Cloud Formation.
To use ecs-service
you need to use a Cloud Formation template
which creates an ECS service along with a Task Definition.
When creating your service, you must supply this template along with the version of the service you wish to run. The tool will create a CF stack with the name you provide.
After the service is created, you can issue commands to update the running version of the service which will result in a stack update.
Instead of having a separate CF template per environment, the idea is to have a single template which works across all environments. Because the tool lets you supply parameters and environment variables to your docker container, your stack template can remain the same between environments.
- NodeJS v6.x or newer
- A Cloud formation template with an ecs service and task definition defined
- Required template parameters:
AppVersion
- Must be defined so the version of the service can be supplied byecs-service
AppDesiredCount
- Must be defined so the number of instances can be supplied byecs-service
- Required template parameters:
- An AWS user/role which has permissions to:
- DescribeStack
- GetTemplate
- UpdateStack
- CreateStack
- DeleteStack
npm install -g ecs-service
AWS REGION
You must export the AWS region you are deploying your service to,
or supply it via the -r
parameter.
export AWS_REGION=us-east-1
AWS Credentials
The tool will use the standard credential chain used by other aws cli tools.
If you wish to supply credentials manually see Options
To create or update an existing service you must provide:
- stackname - The name of the service you wish to create (this will be the name of the CF Stack)
- version
- The version of the service you wish to run. This version should exist in your backing docker repository as a tag.
- If you do not wish to change the running version specify
current
as the version. (Only applicable for existing services)
- template file - A JSON Cloud Formation template containing your ECS Service and Task Definition
- parameter file - A file containing a key value JSON object which maps your template parameter's to their values.
To deploy the service, use the following command:
$ ecs-service deploy [stackname] [version] [template_file] [params_file]
This command will create/update a CF stack with the provided stackname
and supply the service version via the AppVersion
parameter.
For the remaining non-default parameters, it will use the parameter file
.
The command will wait until the successful creation/update of the stack.
If your docker container is configured via Environment Variables you must supply them via the Container Definition's Environment property. This can be tedious to update when your environment variables change. Hardcoding environment specific values in your task definition would require you to maintain several copies of your task definition; one per environment.
ecs-service
allows you to supply a env
file when creating/updating services which
will be used to populate the Container Definition's Environment property.
The --env-file
parameter can be used to supply an env
file containing all of
your environment variables.
Example ENV file
EXTERNAL_SERVICE_URI=http://api.myservice.com
FOO=test
BAR=blah
Each line defines an environment variable.
Example Usage
Using the --env-file
parameter, you can supply a env
file which will be used to pass environment variables to your container.
$ ecs-service deploy [stackname] [version] [template_file] [params_file] --env-file <file>
This parameter can be used with any of the commands provided.
Encrypting Environment Variables
Encrypted environment variables can be supplied to your docker container using kms-env.
If you need to run a new version of your service or update environment variables you can use the simpler run
command.
Instead of having to provide a template and parameter files you can use this command if you only need to adjust the version
or scale of your service.
This command requires:
- stackname - The name of the existing service stack.
- version
- The version label for the service. A corresponding tag should exist in your docker registry.
- If you do not wish to change the running version specify
current
as the version.
$ ecs-service update [stackname] [version]
This command will update your ecs service to use the specified version of your container image.
As mentioned under the Deploy command, you can use the --env-file
parameter to supply a file containing
environment variables for your container.
To stop a service, you can issue the stop
command which will set the desired count of your service to 0.
$ ecs-service stop [stackname]
Note: this will not remove/delete your ECS service or Task Definition.
To completely remove the resources associates with your service stack you can use the destroy
command.
This command will result in a stack deletion operation.
$ ecs-service destroy [stackname]
To specify the number of instances of your service to run, use the --scale
parameter.
Example
To run two instances of an existing service called myservice
we can issue the following command:
$ ecs-service update myservice 0.1.0 --scale 2
If you need your service stack to have associated tags, you can do so via the --tag-file
parameter.
Example
A tag file is a JSON file with an object where the keys are names of tags to create and values are the tag values.
Example Tag File - tags.json
{
"Owner": "Alice",
"Project": "Top Secret"
}
Given the above tag file we can provide it via the --tag-file
command as follows:
$ ecs-service deploy myservice 0.1.0 service.json params.json --tag-file tags.json
The above command will create myservice
using the template in service.json
with the tags Owner=Alice
and Project=Top Secret
To supply credentials manually you can use the following parameters:
- --access-key-id - To specify your access key ID
- --secret-access-key - To specify your secret
- --region - To specify the AWS Region
Using Profiles
Alternatively you can supply an AWS credential profile to use via:
- --profile - To use a credential profile instead of supplying access key and secret