Troubleshooting
Server CPU high: how to identify the workload
A high CPU graph tells you the server is busy, not why. The next step is to identify which processes are using CPU and then connect that activity to requests, jobs, queries, deployments, or unexpected software.
Check whether the spike is brief or sustained
Short bursts during backups, package updates, builds, or traffic peaks can be normal. Sustained saturation that increases request latency or queue depth is more likely to need investigation or capacity changes.
Identify the process, then the trigger
A web worker at 100% CPU is only an intermediate clue. Match it to access logs, slow requests, plugin activity, a specific API endpoint, database query, scheduled task, or application trace.
Include database and I/O wait in the picture
A server can feel CPU-bound when tasks are actually waiting on storage or database locks. Review load, I/O wait, query activity, and process state instead of relying on one percentage.
Rule out abuse and compromised workloads
Unexpected miners, brute-force traffic, vulnerable applications, or compromised accounts can create unexplained compute. If the workload does not match legitimate activity, preserve evidence and investigate security before simply adding CPU.