This PR fixed online upgrade halting issue after sandbox restart. We use vclusterOps to collect node details in online upgrade rather than vsql. This change will remove active connections on the subclusters so subclusters doesn't need to wait for user connections to be paused. As a result, it will fix the halting issue.
This PR fixed online upgrade halting issue after sandbox restart. We use vclusterOps to collect node details in online upgrade rather than vsql. This change will remove active connections on the subclusters so subclusters doesn't need to wait for user connections to be paused. As a result, it will fix the halting issue.