Open berimbolo13 opened 3 years ago
I have the same issue. Also wondering if anyone know how to fix this?
I've encountered this a few times when some topics in our Kafka broker behaved erratically. The solution we had was to delete our RocksDB files and rebuild the files. It's not an ideal fix, but it did fix our environment.
Steps to reproduce
I cannot reproduce it manually but it happens in the production environment sometimes.
We have a single node environment that runs the solution via docker-compose. Kafka cluster contains only one broker. During log compaction inside kafka something went wrong and kafka was trying to remove segment twice then went down for 5 mins, then rebooted:
Faust was restarted automatically but faced the next exception:
Expected behavior
Faust should start successfully.
Actual behavior
Persisted offset in rocksdb is greater than offset in kafka.
Notes
From my understanding, it could happen due to a single Kafka broker:
Did someone face the same issue?