live telemetry · sampled every 2s

mecasasu.casa

A shared 2-core box in kvm, running Ubuntu 26.04 LTS. Everything below is read straight from the kernel — no agent, no polling daemon, no database. Just /proc, a Go binary and an open connection.

dedirock-95892408 192.255.214.140 Intel(R) Core(TM) i9-14900K kernel 7.0.0-14-generic up 12h 29m go1.26.0

Vitals

last sample —

Processor

2 vCPU

1 busy

load 1m: 0.24
load 5m: 0.26
load 15m: 0.23

Memory

3.82 GiB

23 used

in use: 886 MiB
available: 2.95 GiB
cache: 927 MiB

Storage

ext4

13 full

free: 48.5 GiB
used: 7.25 GiB
inodes: 3.0%

Network

—

down —

—

up —

—

Trends

rolling 3-minute window

CPU utilisation

0.8%

per core2 logical

user: —
system: —
io wait: —
steal (noisy neighbours): —

Memory pressure

3.82 GiB

swap — / — · —

truly free: 2.32 GiB
page cache: 927 MiB
open connections: —
processes / threads: — / —

Filesystems

58.8 GiB

Capacity is what a normal user can actually reach — ext4 holds back about 5% for root, and this counts that reserve as unavailable.

Throughput

shared scale

inbound

outbound

What's running

top 8 by CPU

Processes

Command	PID	User	CPU	RSS	Thr	State

Services

Queried from systemd every 10 seconds. A dot goes green when the unit's ActiveState is active.

Spec sheet

what you're working with

Compute

model: Intel(R) Core(TM) i9-14900K
logical cores: 2
clock: 3.19 GHz
architecture: amd64
virtualisation: kvm

Platform

os: Ubuntu 26.04 LTS
kernel: 7.0.0-14-generic
hostname: dedirock-95892408
public ip: 192.255.214.140
booted: 2026-09-05 23:58 UTC

Capacity

memory: 3.82 GiB
swap: 2.00 GiB
disk: 58.8 GiB
filesystem: ext4 on /dev/vda2
go toolchain: go1.26.0

Memory budget — how to spend 3.82 GiB

planning guide

Base OS ~600 MiB
nginx ~40 MiB
PostgreSQL ~768 MiB
Go services ~1 GiB
Page cache ~768 MiB
Headroom ~500 MiB

The real constraint here is RAM, not CPU. A Go service idles around 15–40 MiB, so a dozen of them fit comfortably. What does not fit: a JVM app, Elasticsearch, more than one database engine, or go build on a huge module while Postgres is under load. Two cores also means a build will use both — expect the dashboard to spike while you compile. That is fine.

New here? Start with the handbook.

Shell access, /srv layout, Go projects, just, how nginx actually routes a request, pointing a domain here, free subdomains forever, TLS, databases, passkey auth, and cheap media hosting on Google Cloud Storage.

Open the handbook