I attended this week's meeting and heard a proposal to designate a cluster for user specified distribution scheduling.
I think I may have encountered this demand now. That is to say, the cluster where my Karmada control plane is located cannot access the apiservers of member clusters, and I joined the federation through a pull method. Then during a scheduling process, I encountered a pod that was in a pending state and wanted to perform a second scheduling, but I couldn't install the plugin because I couldn't access the sub cluster's API. At this point, what I want is to take out the task and reschedule it, but refuse to schedule it to the cluster where the previous task did not run.
I attended this week's meeting and heard a proposal to designate a cluster for user specified distribution scheduling. I think I may have encountered this demand now. That is to say, the cluster where my Karmada control plane is located cannot access the apiservers of member clusters, and I joined the federation through a pull method. Then during a scheduling process, I encountered a pod that was in a pending state and wanted to perform a second scheduling, but I couldn't install the plugin because I couldn't access the sub cluster's API. At this point, what I want is to take out the task and reschedule it, but refuse to schedule it to the cluster where the previous task did not run.