Server maintenance is inevitable. Operating system security patches, kernel updates, and hardware maintenance all require downtime. What varies between hosts is how they handle it — specifically, how much notice they give and how long the downtime actually lasts.
Here's what happens when we perform maintenance on the servers your site runs on.
What Triggers Maintenance
We perform maintenance for three categories of change:
Security patches: When a critical vulnerability is disclosed in the Linux kernel, nginx, PHP, PostgreSQL, or other platform components, we patch within 24–72 hours depending on severity. A CVSS score above 8 gets patched immediately.
Kernel updates: We update the kernel on a regular schedule (approximately monthly) to stay current with security fixes and stability improvements. Kernel updates require a reboot.
Hardware maintenance: Occasionally our infrastructure provider requires physical access to the hardware for maintenance. This is scheduled by the provider and communicated to us in advance.
The Maintenance Sequence
When we schedule maintenance:
- Notification sent — we email all customers on the affected server at least 48 hours before planned maintenance. The email includes the start time, expected duration, and what's being updated.
- Status page update — the status page at truecorehosting.com/status.html is updated to show the upcoming maintenance window.
- Maintenance begins — at the scheduled time, we run our pre-maintenance checklist: verify all services are healthy, take a filesystem snapshot, notify the on-call engineer.
- Patch and reboot — the patch is applied and the server reboots. Boot time on our servers is typically 45–90 seconds.
- Service verification — after reboot, our monitoring checks that nginx, PHP-FPM, PostgreSQL, and all customer services are running correctly. We don't close the maintenance window until monitoring confirms clean.
- Notification sent — we send a follow-up email confirming the maintenance is complete and the server is healthy.
Typical Downtime
For a kernel update requiring a reboot: 2–5 minutes from start to services restored.
For security patches that don't require a reboot (application-level patches to nginx, PHP, etc.): 0–30 seconds for a rolling reload.
We don't take extended downtime windows unnecessarily. If a maintenance task can be done in 3 minutes, the window is 3 minutes plus buffer.
Emergency Patches
For zero-day vulnerabilities (actively exploited vulnerabilities with no prior public disclosure), we may patch with minimal notice. In these cases, security takes precedence over scheduling. We notify customers as soon as possible and explain the reason.
What You Can Do
If you need to schedule around our maintenance windows — for example, you have a scheduled marketing campaign or time-sensitive event — contact us and we'll schedule maintenance outside that window.
Maintenance windows are always scheduled outside UK business hours wherever possible.