Skip to content

v2.0.0
Compare
Choose a tag to compare
@JCPistell JCPistell released this 02 Nov 16:44
· 9 commits to main since this release
v2.0.0
3eec89f

Version 2.0 Release Notes

We are very excited to be releasing the next iteration of our reference implementations for self-hosted Looker customers!

This is a major version release which means there are breaking changes

By far the biggest change is we have moved away from our old Kustomize-based approach to Kubernetes deployments and have migrated to using Helm instead. This is based on lots of feedback from you folks and our own growing realization that Helm would better suit our requirements. We think it offers a significantly easier experience.

One really nice feature of using Helm is we can leverage Terraform's Helm provider to better integrate the Kubernetes deployment step straight into the Terraform module... so no more need to deploy the infrastructure first and then follow up with a separate deployment step for the Kubernetes workloads. Simply apply your terraform and you're done!

Beyond the migration to Helm we have a few other nice changes as well:

  • Easier rolling upgrades: Leveraging Helm's pre-upgrade hooks allows us to automatically run the schema migration process that is required for performing rolling upgrades to Looker. Now you simply build a new image, update your Terraform config to point to the new version, apply, and you're done!

  • Easy schedule/render nodes: If your use-case calls for heavy use of scheduling/rendering it may be a good idea to create dedicated nodes for that purpose. Until now this has required a complex set of manual changes - so we went ahead and baked the complexity into the Helm chart. You can now deploy schedule/render nodes by specifying their parameters in your config.

  • Better multi-tenancy: You can now deploy multiple Looker instances into a single Kubernetes cluster/namespace if you need to - this is often desirable for co-locating development or sandbox environments.

Note: Please follow best practices and avoid deploying development and production instances in the same cluster!!

  • Option for automated project creation: You'll find a new Terraform module that will automatically create a sandbox project that is suitable for deploying Looker. We hope you find it useful!

A Note on VM-Based Instances

You may notice that all of these changes focus on Kubernetes-based deployments. At this point we strongly strongly strongly recommend customers adopt a Kubernetes-based architecture for their self-hosted Looker deployments. You will notice we've marked the documentations for VM-based instances as deprecated - we won't be providing any more updates for these reference implementations going forward. Please note that this applies to just this repository - please refer to the official documentation for what forms of customer hosting are officially supported.

Assets 2
Loading