The issue was fixed and everything is back to normal. The impacted customers will be informed.

We've identified the root cause: a recent change to Hog function validation was stricter than intended and caused some valid destinations and transformations to fail. When this happens, matching events aren't forwarded to the destination and those deliveries can't be replayed - the events themselves remain safely in PostHog, but the forwarding for the affected window is lost.

Only a very small number of customers were affected. We've pulled the complete lists of impacted teams and functions in EU and US to start outbound communication.

A fix has been merged that restores the previous behaviour and stops new failures once deployed. We're also scoping a longer-term mechanism to make this class of failure recoverable in the future.

A breaking change in our HogVM caused some destinations to fail. We currently identified the issue and preparing a fix to mitigate the issue.