Reduced Ad Serving and Bidding in U.S. Datacenters
Incident Report for Xandr
Postmortem

Incident Summary
From approximately 14:24-15:33 UTC on Friday, April 29, 2022, Ad Serving was reduced for the LAX1 and NYM2 Datacenters

Scope of impact
During the incident window, clients sending requests from regions near the East and West Coasts of the United States, may have seen increased bid request timeouts causing increases in blanks, defaults, and PSAs

Timeline (UTC)
2022-04-29 14:24: Incident started: an invalid object was created
2022-04-29 14:31: Batches were frozen
2022-04-29 14:33: Invalid object was disabled
2022-04-29 15:08: Hot fix release begun
2022-04-29 15:33: Incident resolved

Cause Analysis
The issue occurred when an invalid object was created using cross-device data without the proper enablement settings

Resolution Steps
Engineering disabled the invalid object and a hot fix was applied to the bidder

Next Steps
• Ensure unit tests cover all cases to catch unusual situations that trigger code issues

Posted May 04, 2022 - 14:20 UTC

Resolved

The incident has been fully resolved. We apologize for the inconvenience this issue may have caused, and thank you for your continued support.

Posted Apr 29, 2022 - 16:00 UTC
Monitoring

We have patched the issue and are monitoring our systems closely. We will provide an update as soon as the issue has been fully resolved.

Posted Apr 29, 2022 - 15:46 UTC
Identified

We have identified the following issue:

  • Component(s): Ad Serving, Bidding
  • Impact(s):
    • Increase in blanks for Console customers
    • Increase in defaults and PSAs for Console customers
    • Some objects may spend under budgets
  • Severity: Major Issue
  • Datacenter(s): LAX1, NYM2

Our engineers are actively working to release a fix to the issue, and we will provide an update as soon as possible. Thank you for your patience.

Posted Apr 29, 2022 - 15:07 UTC