This Month in GKE: July Edition

Key Concepts:

• GKE Alpha Clusters with Feature Gates
• Accelerated Patch Auto Upgrades
• Rollout Sequencing Improvements
• Maintenance Timeline in Upgrade UI
• Extended Support Improvements
• Custom Compute Classes for Node Labels

GKE Alpha Clusters with Feature Gates

• Main Point: GKE now allows more granular control over alpha features in alpha clusters.
• Details: Previously, creating an alpha cluster enabled all Kubernetes alpha features, potentially leading to instability. Now, the gcloud CLI can be used to selectively enable specific alpha features.
• Benefit: Provides more control and reduces the risk of instability associated with enabling all alpha features.

Accelerated Patch Auto Upgrades

• Main Point: GKE can now immediately apply new patch releases to clusters via auto-upgrades.
• Process:
  1. A new patch becomes available in a GKE channel.
  2. GKE immediately sets the auto-upgrade target to that patch.
• Flag: --patch-update=accelerated is used when creating or updating clusters to enable this feature.
• Contrast: Previously, there was a "soak time" before a patch became the default upgrade target. The soak time is still available as an option.

Rollout Sequencing Improvements

• Main Point: Enhanced rollout sequencing feature for cluster upgrades.
• Functionality: Rollout sequencing allows defining stages for cluster upgrades (e.g., dev -> QA -> production).
• Improvements:
  • Increased the maximum number of stages in a sequence from three to four or five.
  • The latest qualified version can be used for subsequent stages, ignoring clusters with higher versions for qualification.
  • Supports a two minor version difference between fleets.
• Example: Upgrading a development environment before a QA environment, and then QA before production.

Maintenance Timeline in Upgrade UI

• Main Point: A visual timeline has been added to the GKE upgrade user interface.
• Functionality: The timeline clearly shows when automatic upgrades can and cannot occur.
• Benefit: Provides better visibility into maintenance policies and scheduled upgrades.
• Feature: Allows review of existing maintenance policies.

Extended Support Improvements

• Main Point: Extended support on GKE clusters now includes more features.
• Extended Support: Offers up to 24 months of support for GKE minor releases.
• New Eligible Features: Clusters using the Gateway API and clusters with Kubelet on the node config are now eligible for extended support.

Custom Compute Classes for Node Labels

• Main Point: Starting with GKE 1.33.2, custom compute classes can be used to define Kubernetes labels on nodes.
• Functionality: Custom compute classes allow users to define Kubernetes labels that are added to the node object in the Kubernetes API.
• Application: These labels can be used with the nodeSelector field to schedule pods on specific nodes.
• GKE Version: Requires GKE version 1.33.2 or later.

Synthesis/Conclusion:

The July GKE updates focus on providing users with more control, visibility, and flexibility in managing their Kubernetes clusters. Key improvements include granular control over alpha features, accelerated patch upgrades, enhanced rollout sequencing, a visual maintenance timeline, extended support for more features, and the ability to use custom compute classes for node labeling. These updates aim to improve stability, streamline upgrade processes, and provide more options for customizing node configurations.