posthog
posthog copied to clipboard
Implement a system for dead letter queue consumption
We need to have a way to consume from the dead letter queue while maximizing for correctness after a full outage.
Changes required:
- [ ] Implement a mechanism to pipe dead letter queue events back through the /capture endpoint
- [ ] Implement a mechanism for determining when we’ve had a full outage and should prioritize dead letter queue events over new events
- [ ] Implement a configurable consumer from the dead letter queue