Process Manager
Process Manager is a live, browser-based top/htop for ROOT: every process on the server in a sortable, filterable table, with CPU and memory bars, the ability to terminate, force-kill or reprioritize a process, a parent-child tree view, and a monitoring tab with live sparklines and top-consumer lists.
Tools → Process Managerhttps://YOUR-SERVER-IP:8443/tools/processes (ROOT only)| PID | Name | User | CPU | Mem | Actions |
|---|---|---|---|---|---|
| 4821 | php-fpm | john | 8.2% | 84 MB | stop · kill · nice |
The summary cards
Four cards head the page: total CPU and Memory usage (as bars), the total process count with a running sub-count, and server uptime with the load average underneath. Above the table, a status summary shows running, sleeping and (only if any exist) zombie process counts.
The process table
Each process is a row with: PID, Name (with its full command underneath, truncated with the whole thing on hover), User, Status (colored: running green, sleeping blue, zombie red, stopped amber), CPU and Memory (each a bar plus the figure, memory also in MB), Priority (the nice value, colored by how aggressive it is), and Runtime. You can sort by PID, name, CPU or memory by clicking the column header (click again to reverse).
Filter the table by a search box (matches name, command, user or PID), a user dropdown, and a status dropdown (running/sleeping/zombie/stopped). A "showing X of Y" counter tells you how much the filter hid.
Actions
Every row carries three actions, each behind a confirmation:
- Terminate sends SIGTERM, the polite "please shut down" signal that lets a process clean up.
- Force Kill sends SIGKILL, the immediate, unconditional kill for a process that will not stop.
- Change priority opens a renice dialog (nice value -20 to 19) to make a process more or less CPU-greedy.
When you filter by a specific user, a Kill All button appears to terminate every one of that user's processes at once, behind a summary confirmation.
Tree view and the monitoring tab
A tree view toggle re-draws the table as a parent-child hierarchy (with expand/collapse and an option to hide kernel threads), which is the way to understand which process spawned which. A second Monitoring tab shows live CPU and memory sparklines (the last several samples) plus top 5 by CPU and top 5 by memory tables, the quickest way to spot what is eating the server.
/proc directly, refreshed by polling on your chosen interval (1 to 30 seconds, with a pause toggle), and it stops polling entirely when the browser tab is not visible to save resources. The list is capped for performance on very busy servers, so on a machine with a very large number of processes the table shows a subset (ordered by PID) rather than every last one.Access
Strictly ROOT, at the route and the backend. See roles and permissions. There is no license gate; the restriction is the role, because ending processes on the host is an owner-level power.