Status → Graphs shows two large live graphs. They are useful to see whether the PBX is overloaded during busy hours, or how much bandwidth the calls use. Point at a graph to see the time and the exact values of that moment in the line below it.
CPU load
The upper graph, CPU load, shows the CPU usage of the whole system in percent (0–100 %, all CPUs together). A plain PBX with a few calls stays in the low single digits. Sustained high values point to transcoding between codecs, many conference participants, or a system that is too small.
Network traffic
The lower graph, Network traffic (all interfaces), shows the incoming (in) and outgoing (out) traffic in bytes per second, summed over all network interfaces except the loopback interface. As a rule of thumb, a call with the G.711 codec (alaw/ulaw) needs about 10 KB/s in each direction per call leg, including packet overhead.
How the graphs are sampled
The PBX does not record these values in the background. They are measured only while a page with graphs is open – this page, or the small graphs on Status → Summary – every two seconds on this page. The PBX keeps the last 120 samples, about four minutes on this page. When you open the page after a while, the graphs start empty and fill up while you watch.
For long-term monitoring, collect the metrics with Prometheus from /api/v1/metrics, see REST API and MCP.