Monitoring and Metrics
ItdaStream collects and exposes comprehensive system and application metrics for monitoring and alerting.
- Metrics collected: CPU, memory, disk I/O, network throughput, per-API TPS/latency, cache hit rate, S3 latencies
- Storage: Time-series in RocksDB, controller aggregates from all brokers every 30s, 7-day retention
- Prometheus:
/metricsendpoint with Gauges, Counters, and Histograms grouped by broker ID - JMX: Automatic JMX Reporter for Java management tools