Infrastructure Environments

1. You are here:
2. Engineering
3. Infrastructure
4. Infrastructure Environments

Maintained by:

On this page

• On this page
• Environments
  • Development
  • Demo
  • .org
  • Review Apps
  • Ops
  • Production
  • Production-Canary
  • Staging
  • Staging-Canary
  • Staging Ref
  • Pre
  • Release
  • GitLab Team Services
  • GitLap
  • Env-Projects
• Self-Managed

Environments

Terraform control for the environments can be found on ops

Development

Development happens on a local machine. Therefore there is no way to provide any SLA. Access is to the individual dev. This could be either EE/CE depending on what the developer is working on.

Demo

This should be a fully featured version of the current EE release. The high SLA and tightened access is to ensure it is always available for sales. There are no features (feature flags/canary/etc) that we do not ship.

.org

Currently there are two main uses for the .org environment:

• Builds.
• Repos that are needed in case GitLab.com is offline.

This is a critical piece of infrastructure that is always growing in size due to build artifacts. There are discussions to make a new build server where nightly CE/EE builds can be deployed or to move the infra repos to a new host that would be an separate (not gitlab.com) EE instance. Although the environment has dev in its domain name don't refer to it as dev since that could be confused with a local development environment.

Review Apps

Ephemeral app environments that are created dynamically every time you push a new branch up to GitLab, and they're automatically deleted when the branch is deleted. Single container with limited access.

Ops

The ops environment hold all infrastructure that is critical for managing GitLab.com infrastructure.

At this time it includes:

• Proxy for ElasticCloud.
• Internal monitoring infrastructure that serves dashboards.gitlab.net
• An isolated GitLab deployment that serves as a backup for all operations related GitLab repositories.
• CICD jobs for critical operations tasks such as backups and maintenance.
• Runners that need to connect to production infrastructure, such as GitLab chatops.

Production

Production will be full scale and size with the ability to have a canary deploy. Production has limited access. It consists of two stages:

• The canary stage is a subset of infrastructure that reaches a limited number of members of the community. We deploy to this stage first. For more information see canary testing.
• The main stage serves the remaining traffic for the wider GitLab community.

Production-Canary

Production-Canary is a environment subset or deployment "stage" in the Production environment, sharing most of the same infrastructure as Production. This additional stage is designed to assist us with rolling out new releases to end users in a more controlled fashion, hoping to catch issues affecting users in a way that minimises impact.

Information on how to access production-canary, use it, and what services it covers is documented in our handbook page on canary stage environments.

Staging

Staging has the same topology as Production and includes the same components, since they share the same terraform configuration.

It has a canary environment similar to production, and new releases are deployed and validated in that environment first before going any further. The Staging-Canary environment has some additional features to take note of when it comes to deployment and usage that are detailed in its own environment entry.

Staging deployments precede production deployments as described in releases, but Staging is deployed a lot more frequently (at least every few hours, given the build is green). This would be a static environment with an pseudonymized production database. The DB is a snapshot of the production DB (updated only often enough to keep migration times to a minimum).

If you need an account to test QA issues assigned to you on Staging, you may already have an account as Production accounts are brought across to Staging. Otherwise, if you require an account to be created, create an issue in the access-request project and assign to your manager for review. Requests for access to database and server environments require the approval of your manager as well as that of one of the Infrastructure managers. The same access-request tracker should be used to request this type of access.

Staging-Canary

Staging-Canary is an environment subset or deployment "stage" in the Staging environment, sharing most of the same infrastructure as Staging. This additional stage is designed to assist us with capturing issues arising due to mixed deployments, where we have multiple versions of one or more components of GitLab that share services such as the database. Information on how to access it, use it, and what services it covers is documented in our handbook page on canary stage environments.

Staging-Canary deployments precede Staging deployments as described in releases, with deployments occuring with the same frequency of Staging. It is important to note that there are two sets of blocking smoke and reliable QA tests that are executed on deployment. One set of tests targets Staging-Canary specifically. The other set targets Staging. Both sets of tests must pass for the Staging-Canary deployment to succeed. This is designed specifically to help flush out issues that occur from mixed version deployment environments. You can determine which environment tests are failing in by examining the Downstream QA pipelines.

Staging Ref

Staging Ref is a sandbox environment used for pre-production testing of the latest Staging Canary code. It is a 10k Cloud Native Hybrid Reference Architecture environment. Staging Ref is being deployed in parallel with Staging Canary using Deployer and GitLab Environment Toolkit. The environment can be destroyed and rebuilt automatically if needed. Initial test data is being populated during deployment. Refer to Staging Ref documentation to learn more.

If you need an account to test QA issues assigned to you on Staging Ref, you can log in using your Production account. Otherwise, if you require an account to be created, create an issue in the access-request project and assign to your manager for review. Requests for access to database and server environments require the approval of your manager and one of the Infrastructure managers. The same access-request tracker should be used to request this type of access.

Pre

The pre environment is an environment used for validating release candidates used to prepare final self-managed releases and production patches. It does not have a full production HA topology or a copy of the production database.

Release

The release environment is an environment used for validating security releases, self-managed final monthly and patch versions. It does not have a full production HA topology or a copy of the production database.

GitLab Team Services

The GitLab Team Services Environment is a group of services for different sites run for GitLab.  It comprises the sites listed above.  These are not controlled in Terraform and seek to dogfood GitLab features such as AutoDevops or GitLab pages.

GitLap

The GitLap environment is an older domain primarily used for support testing. All DNS records under *.dev.gitlap.com and *.do.gitlap.com are controlled via terraform in the dev-resources repository.

The only important system is gitlab-runner-builder.gitlap.com which is used as a CI runner by the gitlab-runner project.

Env-Projects

This environment is used as a genesis project from which all other GCP projects used to support/manage/host gitlab.com are provisioned. No compute resources are present in the project, and it is used solely to provide a mechanism for centrally managing GCP projects, provisioning IAM roles/service accounts for infrastructure deployments within those projects, and controlling which APIs are enabled for each GCP project via Infrastructure as Code (terraform).

Reference: https://ops.gitlab.net/gitlab-com/gitlab-com-infrastructure/-/tree/master/environments/env-projects

Self-Managed

These are environments that are run on-premises by the end-user. We have no influence, access or control of these environments.