In some cases, the Post Process Forwarder gets into a bad state and you need to do some manual cleanup.
(This is if you haven’t changed the default number of instances from the main docker-compose, and if you are in the onpremise directory)
- Stop sentry
docker-compose stop
- Start only Zookeeper and Kafka
docker start sentry_onpremise_kafka_1 sentry_onpremise_zookeeper_1
- Hop into the Kafka instance interactively
docker exec -it sentry_onpremise_kafka_1 /bin/bash
- Receive consumers list
kafka-consumer-groups --bootstrap-server 127.0.0.1:9092 --list
- Get group info
kafka-consumer-groups --bootstrap-server 127.0.0.1:9092 --group snuba-post-processor -describe
- Set the offsets to latest
kafka-consumer-groups --bootstrap-server 127.0.0.1:9092 --group snuba-post-processor --topic events --reset-offsets --to-latest --execute
- Exit the Kafka instance
exit
- Stop Zookeeper and Kafka
docker stop sentry_onpremise_kafka_1 sentry_onpremise_zookeeper_1
- Restart the entire stack
docker-compose up
- Check the logs. You should see the standard retry errors as it starts up, but the continuous restarts should be resolved.