Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesTo upgrade a GKE fleet canary-first, define a custom rollout sequence that orders environment stages, selects a labeled subset of production clusters for the canary, and then includes the remaining production clusters in a catch-all stage. Set a soak period for each stage and validate workloads before the sequence advances. This controls which clusters go when; it does not change how nodes in each pool are replaced.
What GKE rollout sequencing controls
GKE rollout sequencing lets platform teams qualify an eligible upgrade in lower-risk clusters before it reaches more critical environments. Fleets group clusters by environment, while custom stages add finer-grained selection: a stage can target clusters with particular Kubernetes cluster labels, including a canary subset within a production fleet.
As an Amazon Associate I earn from qualifying purchases.
Custom sequencing uses the RolloutSequence and Rollout API objects. The sequence defines the ordered stages and their soak durations. When clusters in the sequence become eligible for an upgrade target from their release channel, GKE creates a rollout and progresses it through those stages.
Keep two separate decisions in view: rollout sequencing determines the order and waiting periods for cluster groups; the node upgrade strategy determines how a node pool replaces its nodes. A sequence does not configure or override that strategy.
#1 Best Overall
How to structure a canary-first sequence
- Group clusters into fleets by environment. For example, use separate fleets for testing, staging, and production. This gives the sequence a progression from lower-risk to higher-risk environments.
- Label the production canary clusters. Add a label such as
canary=trueonly to the production clusters you want in the initial production stage. The documented CLI workflow usesgcloud container clusters updatewith--update-labels. Updating labels overwrites existing labels unless you include the labels you want to retain, so check the current label set before applying an update. - Define ordered stages and soak durations. Configure the sequence with the documented gcloud YAML workflow or a Terraform resource. Label selectors use CEL syntax and begin with
resource.labels. Google Cloud’s example configuration uses a seven-day soak for each of three stages: all clusters in development, labeled production canaries, and then the remaining production clusters. Seven days is an example, not a universal default; choose a duration that gives your team time to observe relevant workload behavior and operational signals. - Add a catch-all stage for each fleet. After a label-selected subset, include a stage without a label selector for that fleet. This ensures that clusters not selected by earlier stages are still covered. If a cluster matches multiple stages in the same sequence, GKE assigns it to the earliest matching stage.
- Observe each stage before relying on progression. Monitor rollout status and validate workloads during the soak. Rollout sequencing does not replace application-specific acceptance criteria, workload health checks, or an operator’s decision about whether a change is safe to continue.
A representative order is testing, staging, labeled production canaries, and the remaining production clusters. The exact stages should reflect how your organization separates risk and verifies changes; the official three-stage example is illustrative rather than a required layout.
How to choose between custom stages and fleet-based sequencing
| Choice | Granularity | Controls | Google Cloud console | Best fit |
|---|---|---|---|---|
| Custom stages | Can select label-matched subsets within a fleet, such as production canaries. | Offers additional rollout controls, including specifying upgrade types and managing rollouts. | Not supported. | Google recommends this model for new environments and sequences that need finer targeting. |
| Earlier fleet-based model | Orders broader fleet groups without the same label-selected subset granularity. | More limited than custom stages. | Supported. | Relevant when console support is required or an existing setup uses the earlier model. |
Google Cloud’s GKE documentation recommends custom stages for new environments. The earlier fleet-based model remains available and is the model supported in the console, but it is more constrained.
How upgrade eligibility and release channels affect progression
A sequence progresses when a version becomes the official automatic-upgrade target for the applicable release channel. Manually upgrading clusters in the first stage does not, by itself, qualify a version or make the sequence advance.
Google recommends keeping clusters in a sequence on the same release channel and minor version so they align on upgrade targets. If channels differ, custom-stage sequencing selects a target from the most conservative channel present in the sequence. That can affect when clusters become eligible, so align channels and versions where possible and confirm the target applicable to the clusters you intend to upgrade.
Before a rollout, check current GKE release notes, known issues, API deprecations, and version skew between control planes and nodes. Minor upgrades introduce more changes than patch upgrades and deserve additional scrutiny. Validate the target in pre-production before it reaches production.
What happens when a stage is delayed or unfinished
Maintenance policies can delay cluster upgrades: GKE starts them only within configured maintenance windows and respects applicable exclusions. A stage can begin soaking after the maximum upgrade time of 30 days even if some cluster upgrades are unfinished. Remaining upgrades may continue during the soak period. Therefore, a stage entering soak is not proof that every cluster in it has completed its upgrade.
Rollout sequencing is not an absolute upgrade freeze. Restricting upgrade types through a sequence or using maintenance exclusions does not prevent mandatory automatic upgrades. Account for that possibility in change planning and operational communications.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →How node upgrade strategy changes capacity and disruption
Sequencing does not change the node pool’s configured replacement strategy. Standard GKE node pools can use surge or blue-green upgrades; Autopilot nodes use surge upgrades. The strategy affects capacity needs and disruption independently of the stage order.
Best Value
| Node upgrade strategy | Replacement behavior | Capacity planning consideration |
|---|---|---|
| Surge | Adds nodes according to the pool’s surge settings while replacing nodes. | May require extra nodes and sufficient Compute Engine quota and resource availability. |
| Blue-green | Creates a temporary parallel node-pool environment before completing replacement. | Temporarily doubles node count, so quota, resources, and reservations need to accommodate the parallel capacity. |
Insufficient quota or resources can make upgrades take longer or fail. Review the selected strategy’s capacity requirements before starting a rollout, including reservations where relevant.
Quick Recap
Operational checks before starting
- Confirm that every intended cluster belongs to the appropriate fleet and that the canary label identifies only the clusters meant for that stage.
- Preserve existing cluster labels when applying updates.
- Check that stages include a catch-all for each fleet and that overlapping selectors will not place a cluster earlier than intended.
- Align release channels and minor versions where practical, then verify the eligible target.
- Review release notes, known issues, API deprecations, and control-plane/node version skew for the target version.
- Choose soak periods based on the workload signals and acceptance checks the team needs to observe; do not treat the example seven-day duration as a platform default.
- Verify quota, resource availability, and reservations against the node pool’s upgrade strategy.
- Monitor rollout status and application health during each soak, accounting for maintenance windows, possible unfinished upgrades, and mandatory automatic upgrades.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
