Event Platform Event Loss (Jul 25 - Jul 29)

Incident Report for Akeneo

Resolved

We have completed our investigation into the Event Platform incident that occurred between July 25, 12:05 PM UTC and July 29, 17:00 UTC.

A new internal mechanism, designed to prevent infinite event loops, failed to correctly initialize during a process restart. This caused the system to mistakenly drop approximately 60% of events (~4M events) for approximately 4 days. The issue has been resolved by disabling the faulty mechanism, and event delivery is now stable.

We want to reassure you that no data was lost on the PIM side, as this issue was isolated to the Event Platform's delivery process.

To help you resynchronize your destination systems and capture all product data changes that occurred, we recommend the following: Pull product data from our REST or GraphQL API. You can filter for products updated after July 25, 2025, 12:05 PM UTC. This will ensure you retrieve all changes that might have been missed.

To prevent this from happening again, we have implemented stricter monitoring and are actively working on a plan to introduce a Dead Letter Queue (DLQ) for event retention. We are also enhancing our error handling to ensure such failures are caught immediately.

Thank you for your patience and understanding.
Posted Jul 25, 2025 - 12:00 CEST