You can view info about the hosts that run your workloads, by going to the
Node Detail page:
Selecting one of them, allows you to view info about the node:
This page displays the same information, as if you ran the
kubectl get node command, in a terminal.
This page also allows you to mark a particular node as unschedulable, so that no additional pods are scheduled on it. This is the same as if you ran the
kubectl cordon <node name> command in a terminal. This can be useful, if you have to conduct maintenance on a host in your cluster.
After updating this setting, the scheduler will not start any new pods on this
Marking a node as unschedulable will not affect the pods already running on that node though. Sometimes, it's useful to drain the node, so that all running
pods are evicted, and the node is marked as unschedulable.
For example, if you need to replace a
node , you could first launch a new one to take it's place. After it has joined the cluster, you can run
kubectl drain <node name> , and all existing pods running in a deployment, would be rescheduled on your new
If you are conducting short-term maintenance (for example, restarting a node to apply a patch), you can set the
unschedulable value on a node back to
false when you are done, and newly launched workloads can now be scheduled on the
node . This has the same result as the
kubectl uncordon <node name> command.