DumbMonit

Simple monitoring from homelabs to mid-sized companies. One container, one IP address to type in, useful graphs and alerts in under a minute.

Work in progress

DumbMonit is under active development; until 1.0 the HTTP API and the data model can still change between minor versions (0.x). The changelog lists every change.

DumbMonit reads the network like a weather bulletin: the home page states the sky in one sentence (“Clear skies.” or “2 advisories, 1 unreachable.”) and lists what needs you, before anything else. Severities follow the meteorological ladder (info → advisory → warning), predictions are forecasts, and maintenance windows are scheduled.

InstallDocker Compose, first start, backups Add your first monitorSNMP, network scan, what happens next AlertingBuilt-in rules, quiet by construction

What it watches

Source	What you get
SNMP v1 / v2c / v3	Switches, routers, NAS, UPS, printers. Seven profiles ship with the product and are applied automatically from the device’s sysObjectID. A network scan adds everything that answers in one go.
Proxmox VE	Nodes, virtual machines and containers, storages, cluster quorum, and the age of the last successful backup per machine.
Proxmox Backup Server	Datastore usage and fill-up forecast, deduplication, age and verification of each machine’s last snapshot, failed tasks, garbage collection.
Synology DSM	Volumes, disks and their SMART health, temperature, load, through the NAS web API.
Linux, macOS, FreeBSD and Windows agent	CPU, memory, disks, network, services, containers and uptime of machines that do not speak SNMP, plus temperatures, disk health (SMART) and ZFS pools where the machine exposes them. One command to install, or a Docker image; in relay mode it probes a remote site’s own monitors over outbound connections only.
Services	HTTP(S), TCP port, DNS, ping and TLS certificate expiry, Uptime Kuma style, with a history bar and availability percentage.
Integration packs	A monitor kind declared in a single YAML file — an HTTP API or a Prometheus /metrics page on the monitor — with its own alert rules, installed without a server release.

Talk to it

• An assistant over MCP — Claude Code, Claude Desktop, ChatGPT, VS Code, Cursor — with 28 tools: a read token only looks (status, monitors, alerts, metrics, agents, containers…), a write token can also act (silence, acknowledge, add a monitor, restart a container, post a status-page incident). Secrets are never returned.
• The HTTP API behind everything the UI does, described at /api/openapi.json (OpenAPI 3.1). Tokens (dmt_…) are scoped read or write, can expire, be restricted to a list of networks, and are rate-limited.

What runs

Container	Role	Footprint
dumbmonit	Collection, API, alerting, web UI, and the embedded VictoriaMetrics for time series	~40 MB RAM + the VictoriaMetrics budget (256 MB by default)

One container, one volume: the server starts VictoriaMetrics from the same image, and configuration and state live in an embedded SQLite database. There is no database container. An external VictoriaMetrics can be used instead (DUMBMONIT_VM_URL).

About the name

DumbMonit was called EzyMonit until September 2026. Commands, environment variables, image names and paths were renamed with it; the old EZYMONIT_* variables and ezym_ agent tokens are still accepted. See Upgrading.

DumbMonit is 100% open source under the MIT license, dependencies included: no feature is held back for a paid edition. DumbMonit is free and will stay free.

DumbMonit is provided “as is”, without warranty of any kind. The author and contributors accept no liability for any damage, undetected failure or unsent alert; validating it in your own environment is up to you.