astroctl_infra_kubernetes_upgrade

astroctl infra kubernetes upgrade

Upgrade cluster Kubernetes version

Synopsis

Upgrade a cluster's Kubernetes version.

This operation is IRREVERSIBLE - the control plane cannot be downgraded. Server enforces: no version skipping, no downgrades.

ROLLING UPDATE STRATEGY

The --max-surge and --max-unavailable flags control how nodes are upgraded:

--max-surge: Extra nodes created DURING upgrade (requires additional capacity) --max-unavailable: Nodes that can be DOWN during upgrade (reduces capacity)

These settings are applied PER NODE GROUP. Example with 3 node groups:

Cluster: my-cluster ├── ng-system: 3 nodes (critical) → maxSurge=1, maxUnavailable=0 ├── ng-app: 10 nodes (general) → maxSurge=2, maxUnavailable=1 └── ng-workers: 20 nodes (batch) → maxSurge=10%, maxUnavailable=10%

UPGRADE PROCESS (each node group upgraded sequentially):

ng-system: Creates 1 extra node, drains old → needs 4 nodes temporarily
ng-app: Creates 2 extra, drains 3 at a time → needs 12 nodes temporarily
ng-workers: Creates 2 extra (10% of 20), drains 2 → needs 22 nodes temporarily

TOTAL CAPACITY NEEDED: max(4, 12, 22) = 22 nodes at peak

RECOMMENDED STRATEGIES

PRODUCTION (default - extra capacity needed): --max-surge 1 --max-unavailable 0 Creates 1 extra node, zero downtime, needs N+1 nodes per group

COST-SENSITIVE (no extra capacity needed): --max-surge 0 --max-unavailable 1 No extra nodes, upgrades in-place, temporary reduced capacity

FAST UPGRADE (more parallelism, more capacity): --max-surge 25% --max-unavailable 0 Creates 25% extra nodes, faster but needs more capacity

For other changes (scaling, provider settings), use: astroctl infra k8s update <cluster> -f update.yaml astroctl infra k8s scale <cluster> <nodegroup> --min X --max Y

astroctl infra kubernetes upgrade <cluster-name> [version] [flags]

Examples

# List available versions for upgrade
astroctl infra k8s upgrade my-cluster --list-versions

# Upgrade to Kubernetes 1.30 (uses production-safe defaults)
astroctl infra k8s upgrade my-cluster 1.30

# Preview upgrade without applying
astroctl infra k8s upgrade my-cluster 1.30 --dry-run

# Generate upgrade readiness report (includes capacity analysis)
astroctl infra k8s upgrade my-cluster 1.30 --generate-report

# Skip confirmation prompt
astroctl infra k8s upgrade my-cluster 1.30 --yes

# PRODUCTION: Safe upgrade with zero downtime (default)
# Creates 1 extra node per group, needs extra capacity
astroctl infra k8s upgrade my-cluster 1.30 --max-surge 1 --max-unavailable 0

# COST-SENSITIVE: No extra capacity needed (slower)
# Upgrades in-place, one node at a time
astroctl infra k8s upgrade my-cluster 1.30 --max-surge 0 --max-unavailable 1

# FAST: Parallel upgrades (needs more capacity)
# Upgrades 25% of nodes at a time
astroctl infra k8s upgrade my-cluster 1.30 --max-surge 25% --max-unavailable 0

# BALANCED: Mix of surge and unavailable
astroctl infra k8s upgrade my-cluster 1.30 --max-surge 1 --max-unavailable 1

Options

      --dry-run                  Preview upgrade without applying changes
      --generate-report          Generate comprehensive upgrade readiness report
  -h, --help                     help for upgrade
      --list-versions            List available Kubernetes versions for this cluster
      --max-surge string         Maximum number of nodes (or percentage like '25%') that can be created above desired count during rolling update
      --max-unavailable string   Maximum number of nodes (or percentage like '25%') that can be unavailable during rolling update
      --output string            Output format for upgrade report (json, yaml, markdown) (default "json")
      --yes                      Skip confirmation prompt