On May 16th, GitLab released GitLab 17. We asked our lead engineer and GitLab Champion Craig Gardener to take a look at the release and walk us through some of the capabilities we can look forward to in the coming year.

In summary, the release is a set-up for some of the great things on the horizon. AI is unsurprisingly a feature of that! However, there are also some major changes to the Verify stage functions and how the runners will register with GitLab which could have a major impact if you don’t act soon.

Below is a summary of the key investment areas and details on the changes to the runners.

AI/ML Efficiencies Across DevSecOps

GitLab is pushing hard to be at the forefront of AI in the development space with the introduction of GitLab Duo, ModelOps, and AI agents – innovations set to redefine the DevSecOps landscape. These capabilities are designed to integrate seamlessly into your workflow, ensuring that AI isn’t just a tool, but a core component of your development process. One of the most interesting areas of the development is to continue to bring AI features to those customers who are using self-managed GitLab instances. Allowing all customers, not just those using the SaaS offering, to use AI features is a key differentiator between one of its biggest competitors GitHub. Key deliverables for FY25 include:

• General Availability (GA) of All Existing AI Features: Ensuring that all AI-driven innovations are accessible.
• GitLab Duo for Offline/Air-Gapped Environments: Extending AI capabilities to secure, isolated networks.
• Expansion Across the Software Development Life Cycle (SDLC): New AI features will emerge, enhancing everything from planning to deployment.

Drive Use Case Adoption to Fully Realise Value

Most people only use a small percentage of the capability of GitLab. GitLab recognises that customers who don’t take full advantage of the platform capabilities within GitLab will not see the full value from their investment. GitLab is working to ensure that customers adopt at least SCM and CI/CD capabilities in their Premium tier and include Security and Governance for those on the Ultimate tier.

Particularly with the falling popularity of Hashicorp and its Vault product it will be interesting to see what the native GitLab Secrets management will look like.

CI/CD Catalog and Components GA: Streamlining continuous integration and delivery processes.

The CI/CD Catalog serves as a centralised hub for developers and organisations to share pre-existing CI/CD components and to discover reusable configurations that others may have already developed. Every component published by users will be part of a public catalogue accessible to all users, regardless of their organisation or project.

This approach promotes cross-organization collaboration, allowing the entire GitLab community to benefit from the wealth of CI components available. It’s a powerful step forward in sharing knowledge among GitLab users, enabling developers to harness the collective expertise of the platform.
Self managed customers will need to build out their catalogue from scratch based on components that are shared by the community. GitLab.com has access to the public catalogue as part of their instance.

Differentiating the DevSecOps Platform

As the consolidation of toolchains becomes more prevalent, GitLab is poised to enhance its enterprise DevSecOps platform by expanding planning capabilities and introducing new Value Stream Capabilities. These are capabilities that will continued to be refined through the version 17 release. Both of these features will bring GitLab closer in line with Jira planning capabilities making a consolidation of Jira into GitLab an increasingly viable option.

• Work Items and Product Analytics: Streamlining project management and analytics.
• Extended Vendor-Specific Application (VSA) Integrations: Enabling richer data integration from third-party sources.

Strengthening GitLab.com Performance

With the increasing shift towards SaaS, GitLab continues to invest in its platforms to ensure they meet the evolving needs of digital transformation without the overhead of self-hosting: Cells is an interesting initiative which would see the sharding of GitLab.com into smaller instances that in the future could support geographically distributed instances of GitLab.

• Cells 1.0 Rollout and Hosted Runners GA: Improving infrastructure and execution environments.
• Enhanced Recovery Point Objectives (RPO)/Recovery Time Objectives (RTO) in GitLab Dedicated: Ensuring high availability and minimal data loss.

Breaking Changes:

Runner Registration Changes

An essential update in Version 17 is the revamped runner registration workflow, in time this will break runner connections that are not updated to support this way of registration.

The changes being made to the GitLab Runner Token Architecture are focused on addressing several key issues with the previous system, enhancing security, traceability, and manageability.

What is being changed?

• Elimination of Registration Token: The new architecture removes the need for a registration token, which was previously used to register runners. This change aims to simplify the registration process and eliminate several security concerns.
• Runner Creation via UI: Runners are now pre-created through the GitLab UI, which involves an authentication token that provides better control and tracking.
• System ID Implementation: A unique system identifier (system ID) will be generated for each runner, enhancing the ability to track and manage runners individually.
• Database Modifications: Changes are being made to the database structure to better support the new runner management system, including new tables and fields that facilitate tracking and management.

Why are these changes necessary?

The previous token-based system had several drawbacks:

• Security Risks: The shared use of a single registration token across multiple runners posed significant security risks, such as unauthorised access and difficulty in token rotation.
• Lack of Traceability: There was no effective way to track who created a runner or to trace the source of a leaked token.
• Management Inefficiencies: The old system required manual intervention for token changes and did not support automatic token expiration, making management cumbersome and error-prone.

What are the benefits?

• Enhanced Security: By linking runners directly to user actions in the UI and requiring authentication tokens, security is significantly improved, reducing the risk of unauthorised access.
• Improved Traceability: With the introduction of system IDs and the ability to track the creation and management of runners through the UI, organisations can now effectively trace and audit runner usage.
• Operational Efficiency: The new system simplifies the management of runners by automating many aspects of the registration and management process, which reduces the administrative burden and improves response to changes.
• Better Resource Management: The restructured database and management processes allow for more efficient use of resources, with clearer insights into runner status and activity.

Overall, these changes aim to streamline the management of GitLab Runners, enhance security and compliance, and provide a more robust and traceable system for managing CI/CD pipelines.

Other Technical Changes

• Postgres 13 is deprecated from GitLab 17. Failing to upgrade to Postgres 14 will break the deployment.
• Unified Approval Rules are no longer supported, in favour of multiple approval rules which provide more flexibility. The automatic migration may leave these rules being more restrictive than desired so this is worth checking post-upgrade if you are not manually migrating the rules prior to the 17 release.

Conclusion

GitLab’s version 17 release is exciting in the foundation that it will lay for some game changing functionality to be added through the year. Star of the show is the inclusion of AI throughout the platform for every use case from Planning, Creating, Securing and Deploying. Something to watch is the changes to the runner registration which will provide benefits but will also require technical teams to undertake some work.

Excited for these upgrades? So are we! If you’re in need of a partner in that process, OTTRA is happy to support you.