I have a StarRocks cluster running on version 3.4.8. It is running in shared data mode with s3 as the object store. When a compute node restarts or sometimes even in other situations, the balance of tablets across compute nodes is sometimes very uneven (see image). This does not always resolve even after dropping and re-adding all the compute nodes. Is there a way to rebalance the tablets to be even across the compute nodes?

https://docs.starrocks.io/docs/administration/management/FE_configuration/#lake_enable_balance_tablets_between_workers manual turn on this configuration. it will be turned on by default since v3.5

Thank you! For some reason the value was set to false in my cluster. Somewhat related question: Is there a way to mitigate the issue of hot tablets / partitions during querying? If queries frequently hit the same partitions or tablets, could those tablets be replicated to multiple compute nodes to improve performance?

not right now.

Are there any plans to introduce a feature like this?