How to read the VICIdial Server Performance Report

The Server Performance Report is the one screen in VICIdial that ignores your campaigns and agents entirely and looks only at the physical box. It answers a single question: is this server keeping up? If your dialer feels sluggish or calls start dropping for no obvious campaign reason, this is where you look first.

Turning it on

This report is not collecting data by default. To start logging, open the server record in the Administration section and set the System Performance setting to Y. Once that is on, VICIdial samples the box every five seconds and stores the numbers. You then pick a date-time range using the YYYY-MM-DD HH:MM:SS format, from one day to another, to view the shift you care about.

Note: this is one of the few reports that cannot be downloaded. If you want a record, take a screenshot before the date range scrolls out of your retention window.

The summary block at the top

Before the graphs, you get a short summary for the chosen shift. Read it in this order:

• Calls handled — how many calls this server processed during the shift, a rough measure of workload.
• Total off-hook minutes — the combined talk and connect time across every line, which tracks closely with how busy your trunks were.
• Average Asterisk channels in use — the typical number of live audio paths at any moment. Every leg of every call, including ringing Concurrent calls, occupies a Channel.
• Average system load and peak system load — the headline health numbers. Average tells you the steady state; peak tells you the worst moment in the window.
• Average USER, SYSTEM and IDLE processor percentages — where the CPU spent its time. USER is your application work, SYSTEM is kernel and network handling, and IDLE is spare headroom. A healthy box keeps a comfortable slice of IDLE.

The graphs

Below the summary, four lines are plotted over the same window: number of processes, system load, number of Asterisk channels, and the USER versus SYSTEM processor split. The point of charting them together is correlation. When system load spikes, you can glance across and see whether channels climbed at the same time (a real call surge) or whether processes ballooned while channel count stayed flat (a runaway process, often a stuck Keepalive or a query that will not finish).

flowchart TD
  A["Load spike on graph"] --> B{"Channels rise too?"}
  B -->|Yes| C["Real call surge"]
  C --> D["Lower dial level or add capacity"]
  B -->|No| E{"Process count rises?"}
  E -->|Yes| F["Runaway process or stuck job"]
  F --> G["Restart the offending service"]
  E -->|No| H["Check SYSTEM CPU and disk IO"]

What the numbers mean for capacity

Read across a full busy shift and the report tells you whether you have room to grow. If peak load is comfortably below your core count and IDLE rarely drops near zero, you can push more lines and a higher Predictive dialing ratio. If peak load is pinned high and IDLE collapses every afternoon, that is your capacity ceiling, not a campaign bug. Reading the load figure correctly matters, so it helps to know exactly what that single number represents — we cover that in VICIdial system load explained.

This report is one piece of a wider routine for watching the box over time. For the full picture of what to track and how often, see our guide to monitoring VICIdial server health and capacity.

Prefer to skip the capacity guesswork entirely? Start a VICIfast trial and get a sized, monitored VICIdial box live in under 40 seconds — performance logging is on from the first boot.