In canary deployments, what is a recommended safety practice?

In canary deployments, what is a recommended safety practice?

• A. Roll back only after manual confirmation
• B. Ignore metrics
• C. Delay rollback arbitrarily
• D. Validate predefined metrics and enable automatic rollback on failure

Answer: (D)

In canary deployments, the safety net is to define objective signals and let them drive automatic rollback. By selecting predefined metrics that reflect user impact—such as error rate, latency, request success, and other SLO-related thresholds—you create a clear, data-driven rule for when to revert. When the canary version behaves within those bounds, traffic can be increased gradually; if the metrics indicate problems, the system automatically rolls back to the previous stable version. This approach minimizes blast radius, speeds recovery, and reduces manual toil because decisions are made quickly and consistently based on observed performance. Ignoring metrics or delaying rollback leaves you exposed to issues longer than necessary, and waiting for manual confirmation slows down response. Automatic rollback on failure with validated metrics provides a reliable safety mechanism that protects users and maintains trust during incremental releases.

In canary deployments, the safety net is to define objective signals and let them drive automatic rollback. By selecting predefined metrics that reflect user impact—such as error rate, latency, request success, and other SLO-related thresholds—you create a clear, data-driven rule for when to revert. When the canary version behaves within those bounds, traffic can be increased gradually; if the metrics indicate problems, the system automatically rolls back to the previous stable version. This approach minimizes blast radius, speeds recovery, and reduces manual toil because decisions are made quickly and consistently based on observed performance.

Ignoring metrics or delaying rollback leaves you exposed to issues longer than necessary, and waiting for manual confirmation slows down response. Automatic rollback on failure with validated metrics provides a reliable safety mechanism that protects users and maintains trust during incremental releases.