All systems operational.
Live uptime for the infrastructure that runs Net Intellect itself: the marketing site, the captive-portal endpoints, the self-hosted Nextcloud and Bitwarden, the monitoring stack. Same kit and same patterns we install for clients.
-
netintellect.co.uk-
-
API endpoint-
-
Captive portal-
How this page works
The numbers above are produced by a small Cloudflare Worker that runs on the same edge network fronting the rest of the site. Every 10 minutes it pings each endpoint, computes up/down state and 30-day uptime percentage, and stores the result in Cloudflare KV. The status page above just reads that JSON. No third-party SaaS, no subscription, no opaque dashboard. The same anti-subscription pattern we apply to client work.
What we monitor here
Three public, customer-facing endpoints. Each is something a prospect or client would care about being reachable:
- Marketing site.
netintellect.co.uk, the front door. Cloudflare Pages, edge-cached, no origin server to fail. - API endpoint. The Cloudflare Pages Function that backs the contact form. If it's up, the form works.
- Captive portal. A live working example of the UniFi captive-portal pattern we install for clients, hosted on a Cloudflare Worker. Same architecture, just our demo instance.
The self-hosted services we use internally (Nextcloud for files, Vaultwarden for passwords, Mautic for marketing automation, the Proxmox cluster underneath) are monitored privately on our own LAN. They don't face the public internet, so there's nothing here for an external monitor to ping. We're happy to walk through them on a call.
Maintenance & auto-updates
Two kinds of maintenance happen on this stack, and both are excluded from the uptime numbers above.
1. Nightly auto-updates (recurring)
We auto-patch our self-hosted services overnight using Watchtower (a maintained Docker container-update tool; we run Nicholas Wilde's fork as the original containrrr project hasn't shipped releases for over a year). Restarts take under a minute, typically too short to register on the 10-minute monitor. As a belt-and-braces measure, Watchtower is pinned to a cron schedule rather than an interval, so it runs at the same time every night regardless of when the host last rebooted.
2. One-off planned work (announced)
Anything bigger (major-version upgrades, hypervisor reboots, NAS firmware, certificate work that needs hands on it) gets posted here at least 48 hours in advance with the affected service(s), the start/end times, and a one-line summary of what's changing. During the window we suspend the affected check in the Worker config; pings stop, no false downtime is recorded, and those minutes are excluded from the uptime calculation.
Upcoming windows: none currently scheduled.
Past incidents
None to report. This section will list every confirmed unplanned downtime event with start/end timestamps and a one-line root cause. Planned maintenance windows are kept in the section above instead. Transparency is the whole point of the page.