Advanced Disciplines

Upgrade Strategy and the Ritual of Continuity

Upgrades are inevitable. The ritual is continuity: the platform changes while service remains intact.

Return to Library Governance Apply

Text

Authored as doctrine; evaluated as operations.

Doctrine

Upgrade strategy is governance over time. You cannot freeze a platform; you can only choose how it evolves.

Kubblai doctrine: upgrades should be routine and boring, not heroic and rare.

Skew and compatibility

Respect version skew rules. Validate add-ons (CNI, CSI, ingress controllers, policy engines) for compatibility. Test CRD conversions and API removals before production.

Most upgrade failures are not in the core—they are in the ecosystem.

Node rotation is the real upgrade

Control plane upgrades matter, but node rotation reveals workload truth: PDB correctness, graceful termination, readiness probes, and data persistence.

If node rotation is painful, your workloads are lying about availability.

Rollback discipline

Define rollback boundaries. Some upgrades are not trivially reversible (etcd schema changes, CRD version shifts). Treat those as planned incidents with explicit risk budgets.

Document the decision points for aborting vs continuing.

Canonical Link

Canonical URL: /library/upgrade-strategy-and-the-ritual-of-continuity