Skip to main content
Migration Notice
We're migrating documentation from the old portal into this one. Some things may look a little different or out of place in the meantime — we know, and we're working to get it right. If something's unclear or doesn't look right, let us know.

Agent Dashboard — Host CPU

What this chart shows​

Host CPU % — the percentage of total CPU capacity consumed by all processes on the agent host, sampled from the JVM via OperatingSystemMXBean.getSystemCpuLoad() and reported on every heartbeat. Heartbeats fire at 30 s base + 0–10 s jitter.

This is host-wide CPU, not just the copy engine process. If the agent host runs other workloads, those are included.

How to read it​

Value rangeInterpretation
0–40%Normal — agent has headroom
40–70%Moderate load — copy jobs running; monitor for sustained pressure
70–90%High — consider reducing copy_threads_per_job or distributing jobs across more agents
>90%Saturated — throughput will degrade; add capacity or shed jobs

A spike that follows job dispatch and returns to baseline is expected behavior. A floor that never drops below 60% indicates background pressure on the host.

Polling interval​

Sampled on every heartbeat (~30 s + 0–10 s jitter). Each chart point represents the host CPU value at that instant, not an average over the interval.

Tips​

  • If Host CPU and JVM CPU diverge significantly, a process other than the copy engine is consuming resources on the host.
  • Correlate Host CPU spikes with the Active Jobs count (visible in the Live Agents tab) to confirm the pressure is copy-engine workload.