Help center

Maintenance

Maintenance is a side-state. A healthy provider can still be in maintenance. Here's how we handle it and how you control the alerts.

← All guides

Scheduled vs ongoing

Every maintenance event has two states:

  • Scheduled: announced but hasn't started yet. The vendor posted a window, we noticed, and we're waiting for it to begin.
  • Ongoing: the window is running right now. The provider may or may not be degraded, most maintenance is invisible to end users, but the vendor announced it anyway.

On the Maintenance tab inside a monitor, each entry is tagged with one or the other so you can tell at a glance.

Why maintenance isn't "down"

Early on we treated maintenance the same as a partial outage. That turned out to be wrong. It triggered red / orange pills on providers that were, for all practical purposes, fine. A maintenance window might be "brief restart of a background worker at 3am UTC"; that shouldn't light up your dashboard like a partial outage.

Now maintenance is a side-state. On the monitors list and monitor detail page, you'll see a status pill (Up, Partial Outage, etc.) plus: if a maintenance window is running, a separate In Maintenance chip next to it. The pill reflects actual service health; the chip just tells you maintenance is active.

Concretely: a provider can be Up · In Maintenance at the same time. The Has issues filter on the monitors list and the Ongoing count on the incidents feed both exclude maintenance, so the numbers line up with real incidents.

Alerts are off by default

Maintenance alerts are off for every new monitor. Most people find them noisy: a 30-provider dashboard generates a lot of "Cloudflare is scheduled to rotate certificates next Tuesday 03:00 UTC"-type pings.

To turn them on for a specific monitor:

  1. Open the monitor from the Monitors list.
  2. Click Configure (top-right).
  3. Go to the Notifications tab.
  4. Tick Maintenance incidents under "Select notification types".
  5. Save.

You can also do the inverse, enable maintenance alerts account-wide in Settings, and then tick them off per monitor where you don't want the noise.

When maintenance alerts are useful

  • Databases and storage: RDS, Cloud SQL, DynamoDB. Planned maintenance here usually means a brief failover, worth knowing about before it happens so you can warn on-call or bump a deploy freeze.
  • Payment providers: Stripe maintenance almost never affects charges, but it can affect reports or webhooks. If your billing team cares, enable it.
  • Your own region: if you've filtered AWS to us-east-1, you probably want to know about maintenance there. You probably don't care about ap-southeast-3.

Viewing maintenance history

In the incidents feed and on each monitor detail page, there's a Maintenance filter chip / tab. That gives you a dedicated view without the noise of real incidents. Useful for retrospectives, change logs, and "did the vendor do a thing last week?" detective work.

Didn't find what you needed? Let us know and we'll add it to the guides.