Blog Engineering Shared Runners use Autoscaling
Published on: April 5, 2016
4 min read Shared Runners use Autoscaling

With the latest release of GitLab Runner 1.1, we've introduced autoscaling to help us meet the growing demand


2022 Update - SaaS Runners has evolved since the time of this blog post. See the up-to-date documentation on the SaaS Runners fleet for Linux, Windows, and Mac.

Not only is Continuous Integration built-in with GitLab CE and EE, but we also offer Shared Runners to run your builds in CI for free on Up until recently, you may have experienced a short wait time as your build got queued for a shared runner. With the latest release of GitLab Runner 1.1, we've introduced autoscaling to help us meet the growing demand, and this is now available on Less waiting, more building!

Scaling the service

Projects hosted in GitLab can have CI tasks defined in their .gitlab-ci.yml files. These tasks are performed by runners which are essentially virtual machines which run your builds in Docker containers. These machines can run any of your builds that are compatible with Docker.

On other platforms, similar functionality is only available with an add-on charge. In GitLab it's free to connect your own runners, and we also began offering free Shared Runners on That means Shared Runners are freely available for projects on, whether they are private or public. However, up until recently users would have noticed their builds would be queued to run as they waited for a shared runner to become available for work.

Today we are extending our offering, enabling the recently announced autoscaling feature. This will reduce the build times and also reduce the time required to allocate a new available machine.

As of today, the Shared Runners for use the new GitLab Runner 1.1. GitLab Runner is configured in autoscaling mode with distributed cache and Docker registry proxy for Docker images.

Using Shared Runners

You will be able to continue using the Shared Runners for testing and deploying your private projects.

The Shared Runners will continue to be used to build your static pages that are served by GitLab Pages.

The machines

All your builds run on Digital Ocean 4GB instances, with CoreOS and the latest Docker Engine installed.

Your builds will always be run on fresh machines. This will effectively eliminate possible security issues, as there is no potential of breaking out of the container.

The tags

All Shared Runners are tagged with shared, docker and linux.

You can use these tags in your .gitlab-ci.yml file to limit which runners are used for specific jobs:

  - shared

  - my_private_runner

The above script will configure GitLab to always run your tests on shared runners, and run deployments only on your specific runner, registered with a my_private_runner tag.

What has changed

Previously, runners were configured to always start the mysql, postgres, redis, and mongodb services. However, we are aware that most of our users don't need to use all (or even any) of these services, and have removed them from the default configuration.

If your builds do require one or more of these services, your builds may start to fail unexpectedly. Modify your .gitlab-ci.yml file to add the services required by your application:

- mysql
- postgres
- redis
- mongodb

  script: run-my-tests

Final configuration

You may be interested what GitLab Runner config.toml looks like. It's really simple!

  name = "docker-auto-scale"
  limit = X
  url = ""
  executor = "docker+machine"
    image = "ruby:2.1"
    privileged = false
    volumes = ["/cache", "/usr/local/bundle/gems"]
    IdleCount = 20
    IdleTime = 1800
    MaxBuilds = 1
    MachineDriver = "digitalocean"
    MachineName = "machine-%s-digital-ocean-4gb"
    MachineOptions = [
    Type = "s3"
    ServerAddress = "IP_TO_OUR_CACHE_SERVER"
    AccessKey = "ACCESS_KEY"
    SecretKey = "ACCESS_SECRET_KEY"
    BucketName = "runner"

The above configuration says that the VM will be used only once, making your builds secure. We will always have 20 machines waiting to pick up a new build. We use Digital Ocean 4GB machine in NYC2, with CoreOS Beta and Docker 1.9.1 installed. The runner is configured to use Docker Hub Registry Mirror and Distributed runners caching.

Happy building!

Live webcast: GitLab CI

Sign up for our webcast on April 14th, which includes an overview and tutorial about using GitLab CI. Join to meet with the GitLab CI team and get your questions answered live!

  • Date: Thursday, April 14, 2016
  • Time: 5pm (17:00) UTC; 12pm EST; 9am PST
  • Register here

Can't make it? Register anyway, and we'll send you a link to watch it later!

We want to hear from you

Enjoyed reading this blog post or have questions or feedback? Share your thoughts by creating a new topic in the GitLab community forum. Share your feedback

Ready to get started?

See what your team could do with a unified DevSecOps Platform.

Get free trial

Find out which plan works best for your team

Learn about pricing

Learn about what GitLab can do for your team

Talk to an expert