Thanks for helping improve node-metrics-agent.
This project runs inside host JVM processes, so contributions should preserve host safety first:
- metric collection must be fail-open
- host/JMX read paths must stay bounded and non-blocking
- queues, maps, caches, and thread pools must have hard limits
- repeated internal logs must be throttled
- new metrics should avoid high-cardinality labels
- config parsing must be defensive and keep the last known-good runtime state on reload failure
Use JDK 21 for builds. The produced bytecode targets Java 11.
./gradlew test
./gradlew checkstyleMain checkstyleTest
./gradlew jacocoTestReportSecurity checks:
./gradlew cyclonedxBom
./gradlew trivyScantrivyScan requires a local trivy installation.
Please include:
- summary of the change
- affected module/path type
- failure behavior
- tests run
- rollback notes for operationally sensitive changes
For telemetry pipeline, config reload, filesystem discovery, or JMX surface changes, include focused tests for failure isolation and saturation behavior.