Purpose
The host machine’s vitals: CPU, memory, disk, filesystem headroom, network, and the
health of the monitoring pipeline itself. All Monitoring pages share a time-range
picker (Last 1h default).
Panels
- Host CPU Utilization (%) — whole machine, not just the Relayer.
- Memory Available (bytes) — falling toward zero predicts trouble before it hits.
- Disk I/O (bytes/s) — read/written.
- Filesystem Headroom (bytes) — free space per filesystem; watch the volume the
Relayer’s data lives on. - Network Throughput (bytes/s) — receive/transmit.
- Bridge Collector Freshness (sec since) — seconds since each internal collector
(hostio, pm2, postgres) last reported. Sustained growth = the metrics bridge is
stuck; the charts elsewhere may be stale, not the services. - Prometheus Targets Up (0/1 per target) — scrape targets (node-exporter,
relayer exporters). A target flat at 0 means its metrics source is down or
unreachable.
Reading it
- Slow transfers + high Host CPU but low Relayer/container CPU → something else on
the machine is eating the CPU. - All charts empty → check Prometheus Targets Up and the monitoring containers
(see Troubleshooting).
Screenshots
