Server monitoring
CPU, memory, network and disk activity, in the console or your own application.
Open your graphs
Choose a server in the console and open Monitoring. Select 1h, 24h or 7d. Graphs update every minute while visible. Hover or use the arrow keys on a graph to inspect a measurement. New rentals need a few minutes before their first samples appear; HV1 history begins when collection starts.
Read measurements through the API
GET /v1/servers/{nodeId}/metrics?range=24h
Authorization: Bearer YOUR_API_KEYThis requires server read permission and ownership. An isolated reseller customer uses the parent's backend key with X-UNode-Customer-Id: CUSTOMER_UUID. Keep the key on your backend, verify the buyer owns your stored order, and proxy only that order's measurements to your frontend.
Draw your own graphs
The response includes UTC Unix-second timestamps, a resolution, availability status and a bounded list of points. CPU is percent of allocated vCPU capacity. Memory values are bytes as observed by the host, including guest caches; they are not application memory usage. Network and disk rates are bytes per second. Multiply network rates by eight for bits per second.
Keep null as missing data and leave gaps in graphs. Do not replace it with zero. Handle collecting, unavailable and unsupported explicitly. Poll no more often than every 30 seconds; normal console updates are once a minute. The API clips history to the current rental and caches internal reads for 30 seconds; customer responses are private and no-store.
Root filesystem space
The optional filesystem field is a latest guest-agent measurement of the single filesystem mounted at /. It is independent of the 1h/24h/7d history range and is not a disk-space history graph.
"filesystem": {
"status": "available",
"scope": "root",
"mountpoint": "/",
"basis": "guest_agent",
"observed_at": 1793550000,
"used_bytes": 2515968000,
"total_bytes": 20601044992
}observed_at is UTC Unix seconds when the host received the report. used_bytes and total_bytes are measured filesystem counters; total is the guest agent's capacity for unprivileged users. Mounts such as /boot, separate data disks and duplicate root mounts are never added together. These numbers can differ from the package's allocated virtual-disk size.
Missing or incompatible guest agents, absent counters, duplicate root mounts, invalid sizes, reports older than 180 seconds and reports preceding this rental return status: "unavailable", with observed_at, used_bytes and total_bytes all null. Treat an absent field on older API versions the same way. Hide the disk-space panel unless this summary is available; do not substitute provisioned disk size or disk I/O. Filesystem availability is separate from the top-level host-history status. UNode performs only the fixed read-only guest-agent filesystem RPC; it does not install software or execute guest shell commands for monitoring.
What the graphs measure
Disk I/O is read/write speed, not filesystem space used. Network rates are activity measurements, not billing-grade monthly transfer totals. The package's disk and bandwidth figures remain plan allocations. See the live reference for the complete schema.