Server monitoring

Is this box healthy? Answered in the tab you are already in.

Standing up Prometheus to answer one question about one server is a week you do not have. Tempest reads the numbers with three ordinary Unix commands over the SSH session you already opened, draws them above the terminal, and never sends them anywhere.

Download Tempest See pricing

Monitoring is a Pro feature. Linux fully supported.

Server list with live health information, on desktop and phone

No agent, no exporter

Three commands you would have run anyway

Switch monitoring on for a host and gauges appear above the terminal once the first sample lands. What runs on your server is not a mystery binary.

top for CPU and memory, df for disk, and /proc/net/dev for network — read-only, no privileges, no daemon. Monitoring docs
Nothing is installed on the server, so there is nothing to uninstall when you are done.
Samples are parsed on your device and shown in your tab. They are never uploaded — not to us, not anywhere. How privacy works
Linux is fully supported; macOS and BSD partially; Windows servers have no /proc to read.

Being told

Scheduled checks that push to your phone

Gauges answer a question you asked. The other half is being told about the thing you were not watching.

Schedule a snippet — a health check, a queue depth, a certificate expiry — and get pushed when the run exits non-zero. Snippets & scheduled runs
Long-running commands notify you when they finish, so the twenty-minute build does not need babysitting.
Connection errors on hosts you mark as critical raise a notification too.
Push between your own devices is end-to-end encrypted — the backend relays ciphertext it cannot read. Push notifications

Where alerts land

The channel your team actually watches

A notification nobody sees is not monitoring. Route them where the conversation already happens.

Telegram, Discord, DingTalk, WeChat Work, Pushover, Bark, email over SMTP, or a custom webhook.
Each channel can be tested from the settings screen rather than by waiting for a real failure.
Channels can be shared with team members so an on-call rota gets the same alerts.

In the session

The numbers next to the thing causing them

A dashboard in another window makes you correlate by hand. This is above the terminal, on the same host, at the same moment.

Watch memory while the deploy runs, in one pane. Split screens
Highlight rules colour the lines in the log that matter while you watch. Output highlighting
The same gauges on Android and iOS for the host you are worried about on the train.
Live gauges above a terminal session in Tempest

Frequently asked questions

Does this replace Prometheus or Grafana?
No, and it is not trying to. There is no time-series history, no alerting rules engine and no fleet view. It answers is this box healthy right now for a server you are already connected to.
What gets installed on my server?
Nothing. Tempest runs top, df and cat /proc/net/dev periodically over the existing SSH session and parses the output locally.
Does my monitoring data leave my machine?
No. Samples are parsed on your device and rendered in your tab; they are not uploaded anywhere.
Can I be alerted when a server goes down?
Through a scheduled check rather than a threshold rule: a snippet that tests what you care about, on a schedule, pushing to your devices when it fails. Connection errors on critical hosts also notify.
Is monitoring in the free plan?
It is part of Pro, along with RDP, VNC, Kubernetes, serial, RCON and the AI agent.

Keep exploring

SSH tunneling Free SSH client SSH client for Android Kubernetes terminal AI terminal

One question, answered in the tab you are already in.

Download Tempest Read the docs