Prysm

Source Code

github.com/cobaltcore-dev/prysm

Prysm is CobaltCore's observability tool for Ceph storage clusters and RADOS Gateway deployments. It provides real-time monitoring, SMART disk health checks, and log compliance analysis - either as a standalone CLI or as part of the metrics pipeline feeding Prometheus.

Architecture

Prysm uses a four-tier architecture to collect and process observability data:

Tier	Role
Remote Producers	Collect data from outside the Ceph cluster - RGW bucket notifications, quota usage, RadosGW statistics
Nearby Producers	Operate within the cluster's network - disk health (SMART), kernel metrics, Ceph logs
NATS	Low-latency messaging backbone routing data between producers and consumers
Consumers	Process data for alerts, analytics, real-time dashboards, and log audit

Output formats

Prysm supports three output modes:

• Console - human-readable output for interactive use
• NATS - publishes metrics to the NATS cluster for downstream consumers
• Prometheus - exposes a /metrics endpoint that Prometheus can scrape

Use cases

• Check Ceph cluster health and OSD SMART attributes across all disks
• Monitor RGW request rates, errors, and per-bucket usage in real time
• Audit log compliance - verify that access logs meet regulatory requirements
• Standalone Prometheus exporter for storage metrics not covered by the Ceph exporter

See also

• Observability - Prometheus
• Storage - Ceph