To ensure high reliability and smooth upgrades for all customers, we follow a staged release process for our pipeline applications. This approach validates releases in real-world environments before making them broadly available.
Release Workflow
1. Private Development & Early Validation
New versions are developed and prepared in private repositories. Once ready, we build and publish Docker images to DockerHub, making them publicly accessible for deployment.
At this stage, the focus is on early validation: ensuring the release behaves as expected in production-grade environments.
2. Controlled Rollout to Managed (CDI) Deployments
The release is first rolled out to our managed Customer Data Infrastructure (CDI) deployments.
This controlled rollout allows us to:
- Validate the release under real operational conditions
- Detect and address any issues early in the deployment cycle
If problems are discovered, they are resolved immediately before the release moves forward. Patch versions can be implemented as needed.
3. Source Availability for Self-Hosted Customers
Once the rollout completes successfully and no blocking issues are identified, the release is made available in our public GitHub repository.
At this point:
Self-hosted customers can access the source code
Documentation and public artifacts are aligned with the released version
At this point:
- Self-hosted customers can access the source code from our public GitHub repositories
- Release notes and documentation are published and aligned with the released version
- Docker images remain publicly available via DockerHub throughout this process
Why We Do This
This staged approach helps us:
Improve release stability for everyone - both managed and self-hosted customers benefit from battle-tested releases
Reduce the risk of breaking changes or incomplete features reaching production environments
Ensure that publicly available releases have already been proven in production at scale
Our goal is not to delay access to the latest releases for Self-hosted customers, but to provide higher-quality, better-tested releases that can be adopted with confidence.
Timeline
The gap between Docker image publication and GitHub source availability is typically managed to minimize disruption while ensuring thorough validation.