By default when using AWS the runner does not need a pool file. It can use environment variables to set up a simple pool in memory. However if you want a more complex configuration or multiple pools follow the below quide.
The pool file sets up the build pools that instantiates hot instances (builds do not wait for an instance to spin up). You can have multiple pools, each with a different configuration or that even use different cloud drivers.
pool.ymlis the default file name.
- Each pool only has one instance type.
- There can be multiple pools. Different pipelines can use the same pool
- You can specify the minimum size of the pool.
- You can specify the maximum size of the pool.
- A pool can only be in one region.
- Changing the pool configuration will mean removing the existing images and restarting the daemon.
- If the pool is empty, it will trigger an adhoc instance.
- For Microsoft Windows pools it is important to set platform. As seen in the windows example below.
Pool file sections
It is recommended when you are naming your pool to use alphanumeric characters and start with a letter. This is to avoid issues with naming on amazon/anka/digitalocean/google.
is (this is the same as plaform in other runners) NB windows support is implemented:
This is where we configure the cloud drivers specific configuration. There are a number of different drivers.
Using a pool file
Below is an example of using a pool file with the docker command. We can use a config folder that contains the necessary configuration files.
The below command creates a container and starts the runner.