> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vms.verolabs.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Probe: process_mon

> Monitor running processes with count, CPU/memory, and per-process detail

## Description

The `process_mon` probe monitors running processes on the host, counting matches and optionally collecting aggregate or per-process CPU/memory detail. Matching is by process name (glob, case-insensitive) with an optional executable path filter.

Requires the `ProcessAdapter` platform adapter.

## Config fields

| Field                | Type     | Required | Default               | Description                                                                |
| -------------------- | -------- | -------- | --------------------- | -------------------------------------------------------------------------- |
| `type`               | `string` | Yes      | —                     | Must be `"process_mon"`                                                    |
| `processName`        | `string` | Yes      | —                     | Process name glob pattern. Case-insensitive. Supports `*`, `?`, `[`        |
| `executablePath`     | `string` | No       | —                     | Full executable path to narrow match (ANDed with processName). Exact match |
| `minCount`           | `int`    | No       | `1`                   | Minimum process count. count \< minCount → `down`                          |
| `maxCount`           | `int`    | No       | —                     | Maximum process count. count > maxCount → `warning`                        |
| `collectCpuMemory`   | `bool`   | No       | `false`               | Collect aggregate CPU% and memory bytes across all matched processes       |
| `collectPerProcess`  | `bool`   | No       | `false`               | Collect per-process detail (CPU, RSS, VMS, threads, FDs, disk I/O, uptime) |
| `collectConnections` | `bool`   | No       | = `collectPerProcess` | Collect open connection count per matched PID. Needs `ConnAdapter`         |
| `maxProcesses`       | `int`    | No       | `20`                  | Cap on per-process detail emission (busiest-by-CPU kept)                   |

## Metrics

| Metric                         | Type    | Unit    | Labels                | Condition                   |
| ------------------------------ | ------- | ------- | --------------------- | --------------------------- |
| `vms.process_mon.running`      | status  | —       | —                     | Always                      |
| `vms.process_mon.count`        | gauge   | —       | —                     | Always                      |
| `vms.process_mon.cpu_percent`  | gauge   | percent | —                     | `collectCpuMemory = true`   |
| `vms.process_mon.memory_bytes` | gauge   | bytes   | —                     | `collectCpuMemory = true`   |
| `vms.process.cpu_percent`      | gauge   | percent | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.memory_rss_bytes` | gauge   | bytes   | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.memory_vms_bytes` | gauge   | bytes   | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.num_threads`      | gauge   | —       | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.num_fds`          | gauge   | —       | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.disk_read_bytes`  | counter | bytes   | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.disk_write_bytes` | counter | bytes   | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.uptime_seconds`   | gauge   | seconds | `pid`, `name`, `user` | `collectPerProcess = true`  |
| `vms.process.num_connections`  | gauge   | —       | `pid`, `name`, `user` | `collectConnections = true` |

## Example config

```json theme={null}
{
  "type": "process_mon",
  "processName": "nginx",
  "minCount": 1,
  "collectCpuMemory": true
}
```

## Notes

* The legacy `commandLinePattern` regex selector has been **removed** — use `processName` glob instead.
* When `collectPerProcess = true` and matches exceed `maxProcesses`, only the busiest processes (highest CPU%) are kept.