Senior · Docker and Kubernetes
Rolling updates and compatibility
How do you deploy without dropping accepted work?
Try explaining this aloud before opening the answer. State your assumptions and trace a boundary case.
Reveal a strong answer
Run multiple healthy instances when required, configure rollout availability, and ensure readiness reflects capacity to accept work. A terminating instance must stop new intake and drain within its grace period.
Keep old and new versions compatible with shared schemas, events and clients during overlap. An expand/migrate/contract change can preserve compatibility where a direct destructive migration cannot.
A successful rollout means the new version passes health and workload checks, not just that containers started. Track errors, latency, queue age and dependency load, and retain a rollback artifact whose data assumptions remain valid.
Common trap
Rolling back an image cannot undo an incompatible database mutation.
Try it
A new version removes a field still read by old Pods during rolling deployment. What is safer?
Worked solution
First deploy readers tolerant of both forms, then write/migrate the new form, and remove old support only after no old readers remain. Test the mixed-version window explicitly.
Change the constraint
How does a PodDisruptionBudget differ from rollout strategy and readiness?
How did you do?
Self-assessment, not a compiler pass or interview score. Explain the reasoning unaided before marking understood.
Not marked. Progress stays on this browser, separate from course completion.
Primary references
Checked 2026-10-20. Use documentation for your pinned runtime, driver and tool version.