If a node lacks an attached data disk, ACK by default stores resources, such as the container runtime, temporary container storage, and container stdout logs, on the system disk. This consumes a significant amount of system disk space and can affect node stability. To prevent this, we recommend that you configure a data disk for nodes when you create a new node pool, or attach a data disk to existing nodes without one.
You can attach a data disk to nodes when you create a new node pool, or attach one to existing nodes by replacing the node pool. After you attach a data disk, ACK automatically mounts the directories for the kubelet and container runtime to the data disk during node initialization.
Attach a data disk to new nodes
Log on to the ACK console. In the Attach Data Disk panel, add a cloud disk in the data disk section. For more information about node pool parameters, see Create and manage a node pool.
Attach a data disk to existing nodes
-
Log on to the ACK console and create a new node pool. For more information about node pool parameters, see Create and manage a node pool.
-
Data Disk: Add the cloud disks that you want to attach.
-
Expected Nodes: Set this to the same number of nodes as the original node pool.
-
-
On the Nodes page, select the nodes in the original node pool and click Drain at the bottom of the page. Select the option to drain all selected nodes simultaneously and read the notes carefully. This operation marks the nodes as unschedulable and gradually migrates applications from the original node pool to the new node pool.
-
On the Node Pools page, perform the following steps in the Actions column for the original node pool to decommission it. For more information about the considerations for deleting nodes, see Delete a node pool.
-
Click Scale and set the number of expected nodes for the original node pool to 0.
-
Choose
> Delete to delete the original node pool.
-