During the period of 15:50 (UTC) to 16:00, PostHog's capture endpoint had a partial outage due to a Kafka outage. This resulted in an elevated rate of 503 status code responses.

During this time we weren't able to capture some events, meaning users might experience inconsistencies on their dashboards for that time period.

We are planning to add a secondary resilient storage as a backup for future occurrences in order to guarantee that we aren't dependent on kafka for data retention

Capture is returning 5xx errors