Metrics Exporters rev 03
The Metrics Exporters Small Update turns the storage server into a Prometheus metrics source. Once installed, the storage server exposes system-level, hardware, and ZFS metrics over HTTP so that any Prometheus-compatible monitoring stack can scrape them and build dashboards and alerts.
Note: Metrics Exporters is delivered as an optional Small Update (the x-metrics-exp module). This article describes revision 03. If your system was updated to a newer revision, refer to the matching Extension:Metrics_Exporters_rev_NN article.
What this module provides
After the Small Update is installed, the storage server publishes two Prometheus metrics endpoints:
- Node Exporter — general system and ZFS metrics, available at:
http://<storage-server-ip>:9100/metrics
- IPMI Exporter — hardware sensor metrics via IPMI, available at:
http://<storage-server-ip>:9290/metrics
Both endpoints use plain HTTP. You can open either URL directly in a browser to verify the exporter is running and inspect the raw metric output.
Available metrics
General system metrics (port 9100)
The Node Exporter endpoint provides standard operating system and hardware metrics, including:
- CPU utilization and load averages
- Memory and swap usage
- Network interface throughput, errors, and packet statistics
- Filesystem capacity and usage
- Disk I/O statistics
ZFS pool metrics (port 9100)
In addition to standard OS metrics, the Node Exporter endpoint includes ZFS-specific metrics:
- Pool capacity, usage, and fragmentation
- Read/write throughput and IOPS per pool
- Checksum, read, and write error counters
- Per-vdev I/O latency distributions and queue depths
- Snapshot age — tracks how old the most recent snapshot is on each dataset, useful for detecting stale or missed snapshots
Hardware sensor metrics (port 9290)
The IPMI Exporter endpoint reports hardware sensor readings from the server's baseboard management controller (BMC), including:
- CPU and system temperatures
- Fan speeds
- Power supply voltages and current draw
- System power consumption
Note: IPMI metrics are only available on hardware platforms that include a working BMC. On virtual machines or hardware without IPMI support, the IPMI Exporter endpoint will return an error or empty metric set.
Connecting Prometheus and Grafana
To integrate with your monitoring stack, add scrape targets to your Prometheus configuration pointing at both endpoints. For example:
| Exporter | Endpoint | Recommended scrape interval |
|---|---|---|
| Node Exporter | http://<storage-server-ip>:9100/metrics |
15–60 seconds |
| IPMI Exporter | http://<storage-server-ip>:9290/metrics |
60 seconds |
Once Prometheus is scraping these endpoints, point Grafana at your Prometheus data source and build dashboards using the collected metrics. The community Grafana dashboard library includes pre-built dashboards for both Node Exporter and IPMI Exporter that can be imported directly.
Known issues and limitations
- IPMI metrics require a working BMC. On hardware without a supported BMC or IPMI interface, the IPMI Exporter endpoint returns no useful data. No configuration change can enable IPMI metrics on unsupported platforms.
- No authentication on metrics endpoints. The HTTP endpoints at ports 9100 and 9290 do not require credentials. Restrict access to these ports at the network or firewall level if the metrics should not be publicly accessible.
- HTTP only; no TLS. Metrics are served over plain HTTP. Do not expose these ports to untrusted networks without a TLS-terminating reverse proxy in front of them.
Changelog
rev 03
- Extended ZFS metrics: added per-vdev I/O latency histograms and queue depth statistics
- Removed Telegraf — metrics collection consolidated to Prometheus exporters only
- Added snapshot age metric for stale-snapshot alerting
- Registered as a standard Small Update with a help link in the small-updates grid
rev 02
- Added Prometheus IPMI Exporter (hardware sensor metrics via BMC)
- Added Telegraf with ZFS pool metrics support
rev 01
- Initial release: Prometheus Node Exporter for general system metrics
For further information, refer to the Prometheus documentation, the Grafana documentation, or contact Open-E support.