ecs

ECS Deployment

Create Task Definition

To deploy the agent on AWS Elastic Container Service (ECS) you must first create the agent task definition. To do this using the web admin console:

1. Go to your ECS web admin console and go to the "Task Definitions" tab.
2. Click on "Create new Task Definition".
3. Select the big "EC2" square and click "Next step". The agent only supports EC2 mode and not Fargate at this time.
4. Scroll to the bottom of the page and click on "Configure via JSON".
5. Paste in the contents of the file signalfx-agent-task.json and click "Save".
6. Click on the "signalfx-agent" container definition under "Container Definitions" and find the section on environment variables.
7. Change the value of the envvar ACCESS_TOKEN to the access token of the SignalFx organization to which you wish to send metrics.
8. Click "Update" and finally "Create" at the bottom of the task definition input form to create the task definition.

You can also do this with the AWS CLI tool by issuing the following command:

aws ecs register-task-definition --cli-input-json file:///path/to/signalfx-agent-task.json

Launching the Agent

The agent is designed to be run as a Daemon service in an EC2 ECS cluster.

To create an agent service from the ECS web admin console:

1. Go to your cluster in the web admin
2. Click on the "Services" tab.
3. Click "Create" at the top of the tab.
4. Select:
   • Launch Type -> EC2
   • Task Definition (Family) -> signalfx-agent
   • Task Definition (Revision) -> 1 (or whatever the latest is in your case)
   • Service Name -> signalfx-agent
   • Service type -> DAEMON
5. Leave everything else at default and click "Next step"
6. Leave everything on this next page at their defaults and click "Next step".
7. Leave everything on this next page at their defaults and click "Next step".
8. Click "Create Service" and the agent should be deployed onto each node in the ECS cluster. You should see infrastructure and docker metrics flowing soon.

Configuration

The main technique for configuring the agent is to have a config file downloaded from the network using curl in the agent container's initialization script. By default it pulls from the config file in our Github repository that provides a basic config that might suffice for basic monitoring cases. If you wish to provide a more complex config file you can set the CONFIG_URL env var in the agent task definition to the URL of the config file. This location must be accessible from the ECS cluster.

The default config supports various environment variable overrides, which you can set in the environment variable section of the agent task definition. See agent.yaml for details (hint: it is the config values that are of the form {"#from": "env:VARNAME"...}).