Master the terminology behind safely reverting risky production deployments.
0 / 5 completed
1 / 5
At standup, a dev proposes reverting a deployment to the previous known-good version after detecting an error spike. What is this action called?
A rollback reverts a system to a previously deployed, known-good version, typically used when a new release introduces a serious regression. It contrasts with a forward-fix, which addresses the issue with a new patch instead of reverting. Rollbacks prioritize speed of recovery over root-cause resolution.
2 / 5
During a design review, the team debates whether to roll back or push a quick patch instead. What is the alternative to rolling back called?
A forward fix (or roll-forward) addresses the problem by shipping a new corrective change rather than reverting to the prior version. Teams choose between rollback and forward-fix based on how quickly a safe patch can be produced versus how straightforward reverting is. This decision is a common judgment call during incidents.
3 / 5
In a code review, a dev notices a database migration accompanied the buggy release, complicating a simple rollback. What is this complication called?
When a release includes a non-reversible migration, such as a schema change incompatible with the old code, rolling back the application code alone may not be safe or sufficient. This is why database changes are often designed to be backward-compatible across at least one release. Migration reversibility is a key consideration in rollback planning.
4 / 5
An incident report shows a rollback was delayed because no automated one-click revert existed. What practice would have reduced the delay?
An automated, pre-tested rollback path built into the deployment pipeline lets a team revert quickly under incident pressure, rather than improvising a manual process. This preparation is a standard reliability practice. Its absence is a common root cause traced back in postmortems about slow incident recovery.
5 / 5
During a PR review, a teammate asks how a rollback differs from disabling a feature via a flag. What is the distinction?
A rollback reverts the entire deployed application to a prior version, while a feature flag can disable just the problematic feature without touching the rest of the deployed code or requiring a redeploy. Flags offer finer-grained, faster mitigation when the issue is isolated to one feature. Choosing between them depends on how contained the problem is.