Dynatrace connection pool alerts: saturation threshold vs acquisition wait time for early exhaustion detection

Goal

I want to configure Dynatrace metric events so my team is warned about intermittent connection pool exhaustion in our Java services before requests start failing, rather than after. OneAgent already reports pool usage for our HikariCP pools, and PurePath shows which requests hold connections.

Constraints and uncertainty

Alerting on absolute saturation (active connections equal to the configured pool maximum) only fires once the pool is already exhausted, so by the time anyone reacts, users have seen timeouts. A derived signal such as connection acquisition wait time, or a percentage-of-maximum threshold, might fire earlier, but I am unsure which of these metrics OneAgent exposes per pool in my version, and whether wait time is available for every supported pool library or only some.

I also need to avoid alert noise: pool sizes differ across services, so a single global threshold seems wrong, but maintaining per-service thresholds adds operational overhead. The environment is Dynatrace SaaS with a recent OneAgent, though exact metric availability may vary by version.

Questions

• Which documented connection pool metrics does OneAgent expose for common Java pools, and is acquisition wait time among them?
• Is a percentage-of-maximum or wait-time-based metric event a reliable early-warning signal compared with saturation alerting?
• What is the recommended way to scope thresholds per service or process group without excessive configuration effort?