v0.9.0 (2022-04-13)

This version introduced static diving scheduling which would help splitting workloads into multiple clusters. For example, if you want to deploy a Deployment with a total of 6 replicas to 2 clusters (“cluster-01” with weight 1, “cluster-02” with weight 2), then “cluster-01” will run such a Deployment with 2 replicas, “cluster-02” runs the other 4 replicas. Please learn more from this tutorial. Several new fields spec.overrides and spec.releaseName were introduced for HelmRelease. Multiple bugs were fixed as well, such as crd storage race condition, in-deleting Description rolling back, etc.

Please use patch version v0.9.1 instead!!!

Changes Since v0.8.0

Full Changelog: https://github.com/clusternet/clusternet/compare/v0.8.0...v0.9.0

What’s Changed

New Features & Enhancements

• Adding a local script that helps establish a local environment with one parent cluster and 3 child clusters. (by @Garrybest in #251)
• Adding tests against golang 1.18. (by @dixudx in #270)
• Introducing new field spec.overrides for HelmRelease. (by @dixudx in #272)
• Introducing new field spec.releaseName (default to the same value as HelmChart name) in HelmRelease, which is backwards compatible. (by @dixudx in #269)
• Using /healthz to report the status of managed clusters whose Kubernetes versions are under v1.16. (by @DanielXLee in #291)
• Introducing static dividing scheduling based on cluster weights. (by @Garrybest in #265, #283, #292, #296 and @dixudx in #285, #289, #286, #295, #297). A new CRD FeedInventory was introduced as well.

Bug Fixes

• Fixing crd storage race condition (by @dixudx in #260)
• Fix resource updates syncing issue (by @DanielXLee in #263)
• Do not rollback in-deleting Description (by @dixudx in #301)
• Fix apiversion assignments with groupversion (by @dixudx in #290)

Misc

• Optimizing ownerReferences with NewControllerRef. (by @dixudx in #294)