You are releasing a new revision of a Cloud Run service. About 5% of traffic should reach it first, then larger shares as long as error rates stay low. Users must not experience any outage. How should you release it?
Show answer and explanation
Answer: D
Cloud Run keeps every revision and lets you split traffic between them by percentage. Deploying a revision with no traffic assignment creates it without sending users to it. You then route a small percentage to it, watch errors and latency, and increase the share until it takes all traffic. The previous revision keeps serving the rest, so the release causes no downtime and can be rolled back by shifting traffic.
Why the other options are wrong
- A. Scaling to zero before deploying causes an outage until new instances start, which violates the no-outage requirement.
- B. A second service with a new name forces a DNS change for every client, creating a cutover window and leaving two endpoints to manage.
- C. Deleting the live revision removes serving capacity before its replacement is ready, so requests fail during the gap.
Exam tip. For canary releases on Cloud Run, deploy with no traffic first and then shift the percentage between revisions.