Staff / expert · Docker and Kubernetes
Stateful databases are not ordinary Pods
Should MongoDB and Redis simply scale like a stateless Go Deployment?
Try explaining this aloud before opening the answer. State your assumptions and trace a boundary case.
Reveal a strong answer
Persistent data needs storage, recovery, replication/failover and backup policies that match the authority of that data. A restarted Pod is not a database recovery strategy. Use supported operators or managed services when their operational contract fits.
Redis persistence and eviction choices differ for a disposable cache versus a queue or authority. MongoDB replication, write/read concerns and transaction support affect behavior during failures. A volume alone is not a tested backup.
Practice restore and failover, measure recovery time/data-loss tolerance, secure network and credentials, and budget client reconnection pressure. StatefulSet provides identity/storage ordering primitives; it does not configure every database correctness property.
Common trap
Three Pods plus a Service do not automatically form a correctly configured database replica set.
Try it
A team says backups are successful but has never restored one. What evidence is still missing?
Worked solution
A restore into an isolated environment, integrity/application checks and measured recovery time. Include keys/configuration needed to decrypt/use the data. Test whether the recovered state meets the promised recovery point.
Change the constraint
What would you throttle when thousands of Go clients reconnect after failover?
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.