Skip to main content
Paygentic runs database maintenance only inside a fixed weekly window: Tuesday, 06:00 to 07:00 UTC. That is 07:00 in Amsterdam in winter and 08:00 in summer. The window is reserved every week, but most weeks nothing happens in it. Maintenance takes place about once every few months.

​
What you may notice

During maintenance, the database is unreachable for a short time, usually under 30 seconds, at some point in the hour. Outside that short drop, the API and the dashboard work as usual. While the database is unreachable:
  • An API request that needs the database returns 503 Service Unavailable with a retry-after header. Wait the number of seconds it gives, then send the request again. See Errors.
  • Usage events are still accepted. The API checks an event against a copy of your event types that it keeps through the drop, then passes it to metering, which does not use the database. An event type or API key that the API has not seen recently can still get a 503. Retry it the same way.
  • The dashboard may show a maintenance message in place of a page that could not load. Reload the page after a few seconds.

​
Advance notice

When maintenance is booked for a window, the dashboard shows a banner from 48 hours before the window starts until an hour after the maintenance starts. For a Tuesday window, the banner appears on Sunday at 06:00 UTC. When nothing is booked, no banner appears.

​
Sandbox and live

Sandbox gets each maintenance update about three weeks before live. If you want to see how your integration handles the short drop, watch for the banner in the sandbox dashboard.