I'm seeing strange behavior when removing CN pods in my shared data cluster running in AKS. My old dev stack was using Starrocks 3.4.1, when I'd scale in a node either with autoscaling or a adjusting the

maxReplilcas

it worked properly. The CN pods would get terminated and they would be removed from starrocks, meaning I would no longer see them in the UI under

compute nodes

. However, in 3.5.2 when I do the same thing, the CNs are not removed from Starrocks. They hang around with this error message

java.net.UnknownHostException: kube-starrocks-cn-3.kube-starrocks-cn-search.kube-starrocks.svc.cluster.local

The only way to clear out the CN is by dropping it from the database via mysql. This breaks autoscaling and frankly seems a bit odd that I need to run a mysql command to resize my cluster. Is this a bug introduced in v3.5.x or working as intended?