Hi. I have cluster starrocks 3.3.18 with FE/BE nodes in shared-nothing mode. I am going to change some configs in my BEs of cluster. (There is no version upgrade) Currently Starrocks cluster used by employer for LOAD jobs (system-load, etc) and query jobs (like select) on all days. I do not want employer see any issue like error in its query/jobs during restart of BE nodes (zero-downtime change). What is the correct approch? I going to use Ansible (infrastructure as code) to do this and I need automatic way (like API, command, etc) not manual human check. I think about some of them but not helpful: • Decommission a node: For a simple rolling change of BEs, lots of data should transfered in whole of cluster and it is not optimize at all. • Blacklist a node: I do not find automatic way to determine whether node after blacklist is not under query or LOAD jobs or anything else. Is there any way? • Script be_stop.sh -g: gracefully stop will wait for running fragments on BE to finish. It seems work for query on BE, but for LOAD jobs which redirected to that BE node, based on my test, LOAD jobs stuck until timeout and failled finally. There are some of ways which I checked. Maybe I do something wrong. Please help me. 🙏