Hi Kafka users, several segments of several topics on more than one of my brokers somewhere have a stretch of about a thousand null bytes.
Comparing with a replica shows that the zeros start after a valid message, and end somewhere in a value. This causes e.g. "CorruptRecordException: Found record size 0 smaller than minimum record overhead". I am suspecting the (virtual) disk. But since I also found index corruption, my question is whether there is a chance Kafka wrote those zeros itself. Maybe a buggy producer? A leader with (other) issues? Thanks Tobias
