Know what is wrong with your servers before your users do.
Code Monitor is lightweight multi-server monitoring written in Rust: one small agent per server, a keyboard-driven dashboard in your terminal, and audit checks that point at real problems, such as a database open to the internet, a certificate about to expire or a unit that keeps failing.
- Open source · AGPL-3.0
- Single binary
- Linux · macOS · Windows
- TLS and mutual TLS
- Prometheus
/metrics
Illustration of the terminal dashboard. Server names and values are examples.
What it watches
The agent (monitor-server) runs its collectors on every server. The dashboard (monitor-client) shows the whole fleet in nine tabs: Overview, Services, Processes, Network, Containers, Postgres, MariaDB, Systemd and TLS.
System
CPU, memory, swap, disks, network interfaces and processes, read straight from the host.
Memory pressure
Not just “memory is high”: kernel pressure stalls (PSI), swap thrashing and memory promised beyond what the host has.
systemd
Failed units, with the usual false alarm of systemd-networkd-wait-online kept at info level.
Docker
Containers, their resource use and health: crash loops, broken health checks, stopped containers, containers with no memory limit and image drift on a tag.
PostgreSQL and MariaDB
Database metrics, idle databases, empty schemas and settings overridden by ALTER SYSTEM.
TLS certificates
Certificates about to expire (warning at 21 days, critical at 7), renewal units that broke, and certificates nothing serves.
Exposed datastores
Database ports listening on every interface, checked against the firewall: critical when public sources get through.
Prometheus and health
/metrics exports every finding in Prometheus format, next to /health and /ready. Bind it to loopback and let your own Prometheus keep the history.
Findings that tell you what to fix
Each audit pass (every 60 seconds by default) produces findings with a severity and a stable target, so an alert names the culprit instead of a percentage.
The dashboard sends alerts to Discord, Slack or any webhook while it runs. For around-the-clock alerting, point Prometheus at /metrics and use your own rules.
- critical
datastore_exposed: a database port accepts connections from the internet. - critical
tls_certificate_expiring: a certificate expires in 7 days or less. - warning
memory_stalled: processes are waiting on memory (PSI). - warning
container_unhealthy: a running container keeps failing its health check. - warning
systemd_unit_failed: a unit failed. - info
image_drift: the image behind a tag changed since the container started.
Install
Prebuilt binaries for Linux (x86_64, aarch64), macOS (Intel, Apple silicon) and Windows are on the releases page.
-
On each server: install the agent
curl -sSL https://github.com/codigocentral/code-monitor/releases/latest/download/install.sh | sudo bash sudo monitor-server show-token # access token for the dashboard -
On your machine: install the dashboard and add the server
curl -sSL https://github.com/codigocentral/code-monitor/releases/latest/download/install.sh \ | sudo bash -s -- --component client monitor-client add --name web-1 --address 203.0.113.10 --token <token> monitor-client # opens the dashboard -
Windows (PowerShell as Administrator)
iwr https://github.com/codigocentral/code-monitor/releases/latest/download/install.ps1 -UseB | iex
The agent listens on port 50051 by default. Enable TLS in the agent configuration, and prefer a private network or VPN between the dashboard and your servers.
How it works
Two programs and nothing else to run: no database server, no web stack, no cloud account.
monitor-client
Your workstation. Terminal dashboard, local history in SQLite, alerts to Discord, Slack and webhooks.
monitor-server
One per server. Collectors for the host, Docker, databases, systemd and TLS; gRPC API; HTTP /health, /ready, /metrics.
Access control
An access token per server, or Ed25519 client keys authorized one by one on the agent. Turn on TLS, or mutual TLS, with certificates from generate-certs.sh.
Your data stays yours
Metrics travel only between your servers and your machine. Nothing is sent to us.
Plans
Everything on this page is free and open source. Paid plans will add what teams ask for: a web dashboard, long history and running it for you.
Community
Free · AGPL-3.0
- Agent, terminal dashboard and all collectors
- Alerts to Discord, Slack and webhooks
- Prometheus
/metrics - Unlimited servers
Pro coming soon
Self-hosted, per server
- Web dashboard with long-term history
- Multiple users, roles and single sign-on
- Reports
Cloud coming soon
Hosted by Código Central, per server
- Everything in Pro, nothing to operate
- Install only the agent