OpenTelemetry-native server monitoring
Every server, accounted for.
Monitorable watches CPU, memory, disks, containers, and services across your fleet — and speaks up only when something needs you. Setup is one command; first charts inside five minutes.
Free for 3 servers·No credit card·Sign in with GitHub or Google
$ curl -sSL https://get.monitorable.io | sh This is the whole setup. The agent registers itself and starts reporting.
Measured on the production fleet we run — updated daily.
How it works
Five minutes, start to finish
No exporters to wire up, no YAML to hand-tune, no proxy of a proxy. Three steps, and the third one is optional.
Run one command
The installer sets up a single lightweight agent — a real OpenTelemetry Collector build — and registers the server to your account. Nothing else to configure.
Watch the data arrive
CPU, memory, disks, network, containers, and services appear on your dashboard within a minute, charted from the first heartbeat.
Decide what wakes you
Turn on ready-made alert rules or set your own thresholds. Monitorable stays quiet until something crosses a line you drew.
Capabilities
Everything a server can tell you
One agent covers the whole machine. No plugins to hunt down, no per-feature add-ons.
Host metrics, live
CPU per core, load, memory, and temperatures — charted in real time, with current, average, max, and P95 on every chart.
Disks, before they fail
Usage and per-device I/O, plus SMART health and SSD wear — so drives get replaced on your schedule, not theirs.
Docker containers
Every container's CPU, memory, and state — detected automatically. Containers that died overnight show up red, not gone.
systemd services
The units you rely on, watched around the clock. A failed service raises its hand before a user does.
Network & IOPS
Throughput per interface and I/O per device, with the spikes preserved — not averaged into fiction.
90 days of history
Zoom from the last hour to the last quarter. Downsampling is step-aware, so the max you see is the max that happened.
The product
A dashboard you can read at 2 a.m.
Five statuses tell you the state of the fleet at a glance. Click through for the whole story — every chart, container, service, and disk for that machine, on one page.
- Stale is not Offline. A server that stopped reporting and a server that's down are different problems. Monitorable never confuses them.
- Honest numbers. Current, average, max, and P95 are computed from exactly the window you're looking at.
- Both themes, first-class. Light for the office, dark for the incident.
The 92% spike from 03:12 is still there at every zoom level. That's the point.
Alerting
Alerts worth waking up for
Monitoring usually fails socially, not technically: too much noise, and everyone mutes the channel. Monitorable is built to stay credible.
- Bounded reminders. Every rule sends a capped number of notifications — you pick, one to five. No infinite repeats.
- Self-cleaning alerts. Removed the container or service that was firing? The alert resolves itself. No ghosts in the queue.
- Restart loops, spotted. A service that fails, restarts, and fails again is one pattern — not five separate incidents.
- Silence is a signal. An agent that stops reporting is the first thing you should know about. Built in, on by default.
Open standards
Built on OpenTelemetry, not around it
The Monitorable agent is a genuine OpenTelemetry Collector distribution — not a proprietary agent with a standards sticker. Your metrics travel as OTLP, the same format the rest of your tooling is converging on.
- Nothing to learn. You don't need to know what OpenTelemetry is to use Monitorable. Install, done.
- Nothing hidden. The agent is open source — read exactly what runs on your servers.
- Nothing kept. Your data exports in open formats. If you ever leave, your instrumentation goes with you.
What the agent can't do
- ✕ No shell access
- ✕ No inbound ports
- ✕ No remote commands
- ✕ No file collection
It reads metrics and ships them. That's the whole job — and you can verify it in the source.
From the field
Engineers are hard to impress
"Setup really was one command. I spent the next hour looking for the catch and didn't find it."
"First monitoring tool in years where the alert channel is still unmuted."
"We retired a Prometheus stack we'd been babysitting for two years. Nobody misses it."
Pricing
Pricing you can do in your head
Flat, per server. Every feature on every plan — alerting, containers, teams. The bill grows when the fleet grows, and never for any other reason.
Free
3 servers · every feature · 14 days of history. For the boxes you care about at home or the project you're just starting.
Start FreeTeam
Unlimited servers · 90 days of history · unlimited teammates. One number, multiplied by your fleet. That's the whole invoice.
Start MonitoringFleet
Volume rates, SSO, and invoicing for large fleets. You'll talk to an engineer, not a quota.
Talk to UsNo usage dimensions. No per-metric math. No surprise line items.
FAQ
Questions engineers ask
How heavy is the agent?
Light. It's a single static binary — an OpenTelemetry Collector build — that typically uses a few tens of megabytes of RAM and negligible CPU. No runtime, no dependency chain.
What happens if the agent stops reporting?
The server is marked Stale — deliberately distinct from Offline — and the built-in agent-offline rule notifies you. A gap in monitoring is information, and it's never silent.
Do containers and systemd services work out of the box?
Yes. The agent detects Docker automatically and watches your systemd units with no extra exporters and no config. Containers that exit stay visible — marked dead, in red — instead of vanishing.
How is this different from Datadog or UptimeRobot?
Uptime checkers tell you the site is down. The big platforms tell you everything, priced in dimensions you'll need a spreadsheet to predict. Monitorable is the middle that was missing: deep server health, flat pricing, five-minute setup.
Which platforms are supported?
Linux — x86-64 and ARM64, on bare metal, VMs, and LXC. That focus is why setup is one command. More platforms are planned.
Is my data isolated?
Yes. Tenancy is enforced on every query, not just at the edges. Sign-in supports GitHub, Google, or email, and teams share dashboards without sharing credentials.
Can I get my data out?
Yes — it's OpenTelemetry end to end, and metrics export in open formats. The only lock-in we believe in is the product being good.
Get started
Start with one server
Install the agent on a box you care about and watch the first charts arrive. If it isn't the easiest monitoring setup you've done, removing it is one command too.