Quiz: Change Management

← Back to quiz index

4 questions

L0 (1 questions)

1. What are the three categories of changes and how do they differ in approval process?

Show answer

Standard changes are pre-approved, low-risk, and repeatable (e.g., scaling replicas). Normal changes require peer review and scheduling (e.g., schema migration). Emergency changes have expedited approval and are used only to restore service during active incidents.

L1 (1 questions)

1. Why should rollback criteria be defined before executing a change, not during the incident?

Show answer

During an incident, stress and time pressure impair judgment. Pre-defined rollback triggers (e.g., error rate > 1%, p99 latency > 500ms, any 5xx from changed service) remove the need for real-time decision-making. They also define who can trigger rollback, the exact commands, and the time window after which rollback may no longer be safe.

L2 (1 questions)

1. Why is Friday afternoon a bad time for production changes, and what is the ideal change window?

Show answer

Friday afternoon means reduced staff over the weekend to detect and respond to delayed failures. Some problems manifest hours later (cache expiry, connection pool exhaustion, slow memory leak). Ideal window: Tuesday through Thursday during business hours when the full team is available. High-risk changes should use late-night maintenance windows when traffic is lowest.

L3 (1 questions)

1. How would you enforce a change freeze at the pipeline level rather than just announcing it in Slack?

Show answer

Add a date-check gate in the CI/CD pipeline that blocks deployments during the freeze window. Example: if the current date falls within the freeze range, the deploy job exits with a clear message and non-zero exit code. Exceptions require incident commander approval and a manual override flag. This is more reliable than a Slack message because it prevents accidental deploys by anyone, including automation.