Question 170
Engineers believe they have resolved this failure: A serverless deployment scales rapidly and each new instance starts dozens of SQL connections, exhausting the database. Which follow-up evidence is most convincing?
Explanation
Correct Answer
Maximum scale tests avoid database connection starvation. The observed state validates the proposed fix: Cap per-instance pool size and total scaling connections, using supported pooling patterns. It addresses this cause: Connection budgets were sized per application instance rather than across autoscaling concurrency.
Incorrect Answers
Even if this observation were recorded, it would support a case where a new firewall rule blocks application TCP sessions to a private SQL instance although ICMP tests appear healthy. It would not verify that maximum scale tests avoid database connection starvation.
The result would be meaningful for a problem where a PostgreSQL client begins rejecting encrypted database sessions after a server certificate rollover. For this question, closure depends on confirming that maximum scale tests avoid database connection starvation.
That result would matter when an application remains broken after Cloud SQL HA failover even though the database's primary endpoint is healthy. It does not establish that maximum scale tests avoid database connection starvation, which is the evidence needed here.