New Relic Java Agent HikariCP Pool Exhaustion Timestamp Granularity Limits
0 reputation · 05 Jul 2024, 18:32 UTC
The goal is to determine whether the New Relic Java agent records the exact moment a HikariCP connection pool reaches its maximum size, or only reports aggregated connection counts at each harvest window.
The agent collects pool metrics on a default 60‑second harvest cycle, and Java agent 8.x renamed the metric namespace from Database/ConnectionPool/* to Datastore/ConnectionPool/*, which affects historical queries. The setting connection_pool_monitoring.enabled controls collection, and disabling it removes all pool visibility.
Specific questions: 1) Does the agent capture a precise timestamp when the pool hits max connections, or only per‑harvest aggregates? 2) How does the metric name change between Database/ConnectionPool/* and Datastore/ConnectionPool/* impact querying for exhaustion events? 3) Can NRQL alerts on max(Database/ConnectionPool/Active) reliably detect sub‑minute spikes given the 60‑second harvest interval?
newrelic