Going live, step by step
From a new account to every server watched, with your team invited.
Prerequisites
You need: • A Linux server with root access, on x86_64 or ARM64 (other CPUs only with your own Node.js 20 or newer), running Ubuntu 20.04+, Debian 11+, RHEL 8+ or compatible (Rocky, AlmaLinux, CentOS Stream), Fedora, Amazon Linux 2023, Arch or Alpine (apk add bash curl first) • Outbound TCP 443 to vigilhold.com (the agent's connection, install and signed updates); outbound TCP 8443 to the Vigilhold node addresses is the agent's fallback; during install, also nodejs.org and your distribution's package mirrors • A Vigilhold account — register at vigilhold.com/register
Optional but recommended: • A staging server separate from production — start there, then expand • A Slack or email address for alert notifications
Step 1 — Register and create your org
Go to /register and create an account. Your first account automatically creates an organization and sets you as Owner.
Pick an org name that matches your company or team.
Step 2 — Deploy to a staging server first
On Hosts, press Add host and Generate install command. Copy the one-line install command:
curl -fsSL https://vigilhold.com/install-agent.sh | bash -s -- --token YOUR_TOKEN --dashboard wss://__DASHBOARD_HOST__/agent-ws
Run this as root on your staging server. The agent installs in under a minute and the server appears on Hosts as soon as it connects.
Verify: the server card shows a green dot and live CPU and memory.
Step 3 — Verify your first signal
Confirm data is flowing:
• Open the server from Hosts: its page shows live CPU, memory and disk, and its Processes tab lists what is running • Go to Alerts: whatever the agent's first checks find appears there
If the server page stays empty, check the agent log:
journalctl -u sentinel-agent -f
(on Alpine: tail -f /var/log/sentinel-agent/agent.log)
Step 4 — Configure notifications
Go to Settings → Notifications and add your email. To use Slack, press Add to Slack in Settings → Integrations (Business plan).
Choose which severities are emailed, and whether they arrive instantly or in an hourly or daily digest; critical alerts still go out during quiet hours. A good start: email critical and warning alerts, and leave info off.
You can tune these after you have a baseline for your fleet's noise level.
Step 5 — Expand to production
Once staging is stable for 24 hours:
- Go to Hosts → Add host — generate a new install command for each production server
- Run the install command on each server (one token per server; tokens expire after 30 days)
- Verify all server cards go green
- Check Alerts for any pre-existing findings — these are real and worth reviewing
If you have many servers, reach out to support@vigilhold.com for bulk provisioning assistance.
Step 6 — Invite your team
Go to Settings → Team and open Invite someone. Invitations are valid for 7 days.
Role guide: • Owner — full access, billing • Admin — full access except billing • Analyst — can view audit log and manage alerts, cannot change settings • Viewer — read-only access to server metrics and alerts
Start with Admin for your primary security engineer and Viewer for stakeholders who need visibility.
Getting support
Support is by email, on every plan:
• Email: support@vigilhold.com • First reply within one business day (Monday to Friday, US Mountain Time)
If you think a server is under attack, or Vigilhold did something you didn't expect, put URGENT in the subject: urgent messages get a first reply within 4 hours, any day of the week.
Tips
- Always start with staging. It lets you see real signals without risking noise fatigue on production.
- The first alert after install is often a pre-existing finding — don't panic, just review it.
- Set up notifications before you expand to production so you are paged when something happens.
- If your server is behind a strict firewall, allow outbound TCP 443 to vigilhold.com. Allow outbound TCP 8443 to the Vigilhold nodes too if you can: the agent falls back to it if vigilhold.com is unreachable. During install it also needs nodejs.org and your distribution's package mirrors.
- Tokens are single-use and expire after 30 days. Generate one per server, not one for all servers.