Question 890
Engineers believe they have resolved this failure: A multi-environment deployment applies a production connection string to a staging service and staging tests mutate live records. Which follow-up evidence is most convincing?
Explanation
Correct Answer
Staging requests cannot access production database endpoints or data. This distinguishes actual recovery from a merely healthy-looking console. The targeted correction was: Use environment-specific protected configuration and validate destination identities predeployment.
Incorrect Answers
Even if this observation were recorded, it would support a case where a newly provisioned Cloud SQL fleet has CPU alarms but no alerts for storage exhaustion or replica delay. It would not verify that staging requests cannot access production database endpoints or data.
Even if this observation were recorded, it would support a case where an AlloyDB read pool is scaled up but latency does not improve because one analytical query still performs expensive large joins. It would not verify that staging requests cannot access production database endpoints or data.
That result would matter when a database initialization script retries failed CREATE TABLE statements indefinitely and repeatedly hits service quotas. It does not establish that staging requests cannot access production database endpoints or data, which is the evidence needed here.