The cluster function of Doris is planned to be offline. Doris currently has plans to take the cluster function offline. After going offline, Doris will no longer support the cluster function. There are two main reasons: 1. The cluster function is actually a semi-finished product, and there are many new functions that do not support cluster anymore. 2. The cluster function strongly relies on connection, and many asynchronous jobs in Doris do not have connection information. If an asynchronous job wants to follow the normal planning logic, it must find a way to bypass the connection, which makes the code very confusing.
Now I mainly want to know 1. If you are using the cluster function, what are the main usage scenarios, and are there any risks after going offline? 2. If there is a smooth migration solution, what do you want it to be?