Installing GitLab by using Helm
- Tier: Free, Premium, Ultimate
- Offering: GitLab Self-Managed
Install GitLab on Kubernetes by using the cloud native GitLab Helm chart.
Assuming you already have the prerequisites installed and configured,
you can deploy GitLab with the helm command.
The GitLab Helm chart requires external PostgreSQL, Redis, and object storage. For production, see plan your environment.
For a production deployment, you should have strong working knowledge of Kubernetes. This method of deployment has different management, observability, and concepts than traditional deployments.
If you don’t need Kubernetes for your GitLab instance, see the other installation methods.
Container images
The GitLab Helm chart uses the Cloud Native GitLab (CNG) container images to deploy GitLab.
External PostgreSQL, Redis, and object storage are required. Refer to the following documentation for instructions on how to configure the chart to use external services.
Configure the Helm chart to use external stateful data
Configure the chart to use your external PostgreSQL, Redis, and object storage services, and external Gitaly if you run Gitaly outside the cluster.
Plan your environment
Before you deploy to production, plan your environment’s size and components:
- Assess and plan deployment size to characterize your workload.
- Installation requirements for component requirements, including the supported options for PostgreSQL and Redis.
- Cloud services for supported managed services.
For example environment designs, see the reference architectures.
Use Infrastructure as Code (IaC) and builder resources
GitLab develops Infrastructure as Code that is capable of configuring the combination of Helm charts and supplemental cloud infrastructure:
- GitLab Environment Toolkit IaC.
- Implementation pattern: Provision GitLab cloud native hybrid on AWS EKS: This resource provides a Bill of Materials tested with the GitLab Performance Toolkit, and uses the AWS Cost Calculator for budgeting.
Zero downtime upgrades
Zero-downtime upgrades allow you to upgrade GitLab without service interruptions. To enable this capability, you must configure rolling update strategies during your initial installation. If you add these settings to an existing deployment later, it will trigger pod restarts and may cause brief downtime.
Achieving zero downtime as part of an upgrade is notably difficult for any distributed application. The documentation has been tested as given against our HA reference architectures and resulted in effectively no observable downtime. However, be aware your mileage may vary dependent on the specific system makeup.
During the upgrade, users may temporarily experience UI inconsistencies or HTTP 404 errors for assets as requests route between pods running different versions, these issues typically resolve with a page refresh.
To configure your deployment for zero-downtime upgrades, be sure to include the rolling update settings from the Recommended deployment settings.
For complete upgrade procedures, see the Upgrade with zero downtime documentation.