New Content Cards Not Being Served in EU 01

Incident Report for Braze, Inc.

Postmortem

At 5:36pm ET on August 26, 2020 Braze Engineering was performing maintenance on the database schema underlying Content Cards in EU 01. As part of this maintenance, some unused database columns were dropped. The maintenance plan involved removing and recreating a database view which previously used these columns. The view was incorrectly recreated which caused the database query planner to change the indexes used to respond to queries, which impacted Content Cards sending performance in EU 01 severely. Braze engineering paused the sending of Content Cards on EU 01 from 6:57pm ET until 9:45pm ET while we identified and debugged the issue of correcting the view. Content Cards sending was fully restored at 9:50pm ET.

This maintenance had been run on staging as well as other smaller production environments successfully prior to this change. This was not caught in other environments because the lesser Card volume there did not cause the impacted query to meaningfully change its performance. We apologize for this issue and are implementing tests to ensure any other database schema updates don’t impact content cards sending in this manner.

Posted Aug 26, 2020 - 21:55 EDT

Resolved

This incident has been resolved and all affected Content Cards on EU 01 are now available on end users' devices.
Posted Aug 26, 2020 - 21:55 EDT

Identified

We have identified the issue and our teams are working on a fix.
Posted Aug 26, 2020 - 21:29 EDT

Update

We are continuing to troubleshoot this issue.
Posted Aug 26, 2020 - 19:27 EDT

Investigating

We are currently investigating an issue on EU 01 that is preventing new Content Cards from being served to client devices. No Content Cards intended for an end user are lost, and any existing Content Cards on an end user's device will remain viewable by the end user.
Posted Aug 26, 2020 - 18:13 EDT
This incident affected: EU 01 Cluster (Outbound Messaging).