we are running logstash-6.1.3 with kafka 1.1 in a test environment. Two logstash instances and four brokers.
Yesterday after all machines where suddenly restarted (yes, in test this happens), everything first looked fine but then we saw that when logstash was started, it had read everything from beginning from all topics instead of the latest offset.
I can see in our config that same group and client id´s are used and kafka-consumer-groups show that the same consumer-group + client-id is used in kafka.
Note that this error occured on logstash only, we are using kafka-connect as well without this problem.
I understand that this is a longshot, but any input is appreciated.