Server Metrics Reference
The agent collects the following metrics every 15 seconds (configurable).
CPU
| Metric | Unit | Description |
|---|---|---|
cpu.usage_percent | % | Total CPU utilisation (all cores averaged) |
cpu.user_percent | % | User-space CPU time |
cpu.system_percent | % | Kernel CPU time |
cpu.iowait_percent | % | Time waiting on I/O |
cpu.steal_percent | % | CPU stolen by hypervisor (VMs) |
cpu.load_1 | — | 1-minute load average |
cpu.load_5 | — | 5-minute load average |
cpu.load_15 | — | 15-minute load average |
Default alert thresholds:
cpu.usage_percent > 85%for 5 minutes → Warningcpu.usage_percent > 95%for 2 minutes → Critical
Memory
| Metric | Unit | Description |
|---|---|---|
mem.total_bytes | bytes | Total physical RAM |
mem.used_bytes | bytes | Used RAM (excluding buffers/cache) |
mem.free_bytes | bytes | Free RAM |
mem.used_percent | % | Used / Total |
mem.buffers_bytes | bytes | Kernel buffers |
mem.cached_bytes | bytes | Disk page cache |
swap.total_bytes | bytes | Total swap |
swap.used_bytes | bytes | Used swap |
swap.used_percent | % | Swap usage |
Disk
| Metric | Unit | Description |
|---|---|---|
disk.{mount}.total_bytes | bytes | Total capacity for each mount point |
disk.{mount}.used_bytes | bytes | Used space |
disk.{mount}.free_bytes | bytes | Available space |
disk.{mount}.used_percent | % | Used / Total |
disk.{mount}.inode_used_percent | % | Inode usage (critical for mail servers) |
Default alert thresholds:
disk./.used_percent > 85%→ Warningdisk./.used_percent > 95%→ Critical
Network
| Metric | Unit | Description |
|---|---|---|
net.{iface}.bytes_recv_rate | bytes/s | Inbound throughput |
net.{iface}.bytes_sent_rate | bytes/s | Outbound throughput |
net.{iface}.packets_recv_rate | pkt/s | Inbound packet rate |
net.{iface}.packets_sent_rate | pkt/s | Outbound packet rate |
net.{iface}.err_in_rate | errors/s | Inbound errors |
net.{iface}.err_out_rate | errors/s | Outbound errors |
net.{iface}.drop_in_rate | drops/s | Dropped inbound packets |
Processes
The top-20 processes by CPU and memory are reported every 60 seconds:
{
"pid": 1234,
"name": "php-fpm",
"cpu_percent": 12.4,
"mem_rss_bytes": 134217728,
"status": "running",
"user": "www-data",
"cmdline": "php-fpm: worker process"
}System info (reported once on connect)
- OS name, version, kernel
- CPU model, core count
- Total RAM
- Hostname
- Public IP
- Uptime