Hello, I am investigating a case in which I have a tx log file of about 4GB, the ZooKeeper server has 4GB of memory. The overall database in memory is very small, less than 1 GB.
It looks like that when the ZK server starts and replays the txlog it is using more than 4GB and it crashes. The system is on ZK 3.5.x (it is actually Apache Pulsar 2.7.x). I am not able to reproduce the issue with synthetic data, has anyone ever observed this behaviour ? Enrico
