Postmortem — Dashboard outage and delayed analytics (2026-11-11)
Summary
On November 11, 2026 between 17:38 and 18:02 UTC, the Adapty Dashboard was unavailable for about 21 minutes and dashboard analytics stopped updating for about 24 minutes. A planned hardware maintenance by our hosting provider was applied to the wrong server and took a production host in our US region offline. We moved the affected services to healthy capacity and restored them; the SDK API kept serving profile, paywall and purchase requests throughout.
Impact
-
Duration: about 24 minutes (17:38 → 18:02 UTC); the Dashboard was unavailable 17:38 → about 17:59 UTC.
-
Surfaces affected: the Dashboard (app.adapty.io) and dashboard analytics.
-
Analytics events: between 17:48 and 17:53 UTC part of the SDK analytics events (paywall, onboarding and flow views) sent by apps were not recorded. Conversion charts for that window may show lower counts. Purchases, renewals, profiles and paywall delivery were not affected.
-
Recovery: dashboard analytics caught up automatically once the service was restored; no action is needed from you.
Timeline (UTC)
Time
Event
17:38
A hosting-provider maintenance takes a production server in the US region offline.
17:38
Dashboard becomes unavailable; dashboard analytics stop updating.
17:48–17:53
Part of SDK analytics events are not recorded.
~17:59
Dashboard restored.
18:02
Analytics processing restored and caught up.
Root cause and contributing factors
Root cause: a maintenance request intended fortached to one of our production servers, and the
hosting provider performed the maintenance on it.
Contributing factors:
- Several redundant components of the affected sesame physical host.
- Recovery of stateful components on the lost hos
- Failures to record analytics events were not su time.
What went well
- The SDK API kept serving profile, paywall and pu whole incident.
- Our external monitoring detected the Dashboard o
What we will do
Prevent
- Require explicit confirmation of the target servntenance.
Detect
- Alert on failures to record SDK analytics events
Mitigate
- Spread redundant components across physical hostfor the loss of one host.
We apologize for the disruption. If you see missinwindow, please reach out to your usual Adaptycontact and reference the 2026-11-11 incident.