This demo shows how to do message replay after a failed database.
- Install all components using
kubectl apply -f replay/ - Check the log files of the loader to see that it loaded all the data into database
- Check the database to see that the data are in the table
- Delete the database pod using
kubectl delete pod $(kubectl get pod -l component=postgres -o=jsonpath='{.items[0].metadata.name}') - Check that the database is empty :-/
- Try to restart the loader pod to see that it will not load the messages again (
kubectl delete pod $(kubectl get pod -l component=loader -o=jsonpath='{.items[0].metadata.name}')) - Check the consumer offsets using
replay/describe-group.sh. Notice the current offset column. - Scale down the loader to 0 replicas. If it runs, we cannot reset the offset.
- Reset the offset to 0 using
replay/reset-group.sh - Check the offset again to see how it changed
- Scale the loader up again
- Check the database which now has the records again