Collected metrics
What it is
Observatory collects metrics through ping (ICMP), SNMP polling and HTTP checks. Time-series data lives in a dedicated metrics engine with 180 days of retention; configuration and alerts live in the main database. This page lists what is collected, how often, and how far back you can look.
Metrics collected
Heartbeat (Ping)
| Metric | Unit | Description |
|---|---|---|
| Latency | ms | Round-trip time of the ICMP packet |
| Packet loss | % | Probes sent that got no reply |
| Status | Up / Down | Whether the device answered |
From these, Observatory derives average, minimum, maximum latency and uptime % for the selected period.
Bandwidth and errors (SNMP)
| Metric | Unit | Description |
|---|---|---|
| Inbound traffic | Mbps | Data received on the monitored interface |
| Outbound traffic | Mbps | Data sent on the monitored interface |
| Aggregate | Mbps | In + Out, optional line (Agg button) |
| Interface errors | per minute | CRC errors, malformed frames |
| Discards | per minute | Packets dropped by the interface |
Both 32-bit and 64-bit (high-capacity) SNMP counters are supported, so multi-gigabit links read correctly. You pick the monitored interface from the dropdown in the Bandwidth (SNMP) section.
Resources and environment (vendor OIDs)
On devices where Deep Discovery identified the vendor, Observatory also polls CPU usage, memory usage and temperature. These charts appear automatically once the OIDs are known — see [[crearack—monitoring—deep-discovery]].
HTTP checks
HTTP health checks record status code, response time (ms), SSL validity and days until certificate expiration. Results feed alerts rather than a dedicated chart — see [[crearack—monitoring—ping-http]].
Wireless access points and UPS devices have richer, specialized metric sets in their own modules: [[crearack—monitoring—que-es-wireless]] and [[crearack—monitoring—que-es-ups]].
How often
Every monitored device has its own Polling cadence — 15 s, 30 s, 1 min (default) or 5 min — set in its card editor; its charts refresh at the same pace. See [[crearack—network—device-page]].
| Mode | Check | Interval |
|---|---|---|
| Cloud Only | All enabled checks | While the device tab is open, at the device’s polling cadence |
| Sentinel (Local Agent) | Ping | 24/7 at the cadence — but never slower than every 30 s, so outages are always caught fast |
| Sentinel (Local Agent) | SNMP bandwidth and errors | 24/7 at the cadence |
| Sentinel (Local Agent) | CPU / memory / temperature (vendor OIDs) | 24/7 at the cadence — but never more often than every 60 s (heavier queries) |
In Sentinel mode the Agent batches results and pushes them to the cloud, so charts keep filling even when nobody has Observatory open. The push pace follows your fastest device.
Where it is stored
- Time-series engine (VictoriaMetrics) — all numeric series: latency, loss, bandwidth, errors, CPU, memory, temperature, HTTP timings. Retention: 180 days. Every sample is labeled with your organization, so data is fully isolated per tenant. Writes are asynchronous — charts never wait on storage.
- Main database (PostgreSQL) — target configuration, alert rules, alert event history and recent raw samples.
Maintenance is automatic: recent database samples are aggregated into hourly buckets, raw samples older than 30 days are purged, and resolved alerts are removed after 90 days. The 180-day series in the time-series engine are unaffected.
Time ranges
- Chart range buttons go from 01m up to 7 days (click a button repeatedly to extend) — see [[crearack—monitoring—dashboards]].
- For anything older, use Report mode on a device tab: pick From and To dates and click Apply. You can query any window inside the 180-day retention.
- Zoom into any chart with the slider or mouse wheel to inspect short events.
Related
- [[crearack—monitoring—que-es-observatory]] — module overview
- [[crearack—monitoring—dashboards]] — where these metrics are charted
- [[crearack—monitoring—configurar-snmp]] — enabling SNMP collection
- [[crearack—monitoring—ping-http]] — ping and HTTP checks in detail
- [[crearack—monitoring—alertas]] — thresholds on these metrics
- [[crearack—monitoring—deep-discovery]] — unlocking CPU/memory/temperature
- [[crearack—network—device-page]] — setting the polling cadence per device
- [[crearack—monitoring—que-es-wireless]] — wireless-specific metrics
- [[crearack—monitoring—que-es-ups]] — UPS-specific metrics
Véase también
- [[crearack—monitoring—que-es-observatory]]
- [[crearack—monitoring—dashboards]]
- [[crearack—monitoring—configurar-snmp]]
- [[crearack—monitoring—ping-http]]
- [[crearack—monitoring—alertas]]
- [[crearack—monitoring—deep-discovery]]
- [[crearack—monitoring—que-es-wireless]]
- [[crearack—monitoring—que-es-ups]]