After the configuration of a node pool is updated, some configurations cannot be automatically synchronized for existing nodes. You can manually synchronize configurations for these nodes.
- Do not delete or reset nodes during batch synchronization. Otherwise, the synchronization of node pool configuration may fail.
- This operation involves resetting nodes. Workload services running on the nodes may be interrupted due to single-instance deployment or insufficient schedulable resources. Evaluate the upgrade risks and perform the upgrade during off-peak hours. You can also specify a disruption budget for your application) to ensure the availability of key services during the upgrade.
- During configuration synchronization for existing nodes, the system disk and data disk will be cleared. Back up important data before the synchronization.
- Only some node pool parameters can be synchronized by resetting nodes. The constraints are as follows:
- When editing the resource tags of the node pool. The modified configuration takes effect only for new nodes. To synchronize the configuration to the existing nodes, you need to manually reset the existing nodes.
- Updates of kubernetes labels and taints are automatically synchronized to existing nodes. You do not need to reset nodes.
Synchronizing a Single Node
- Log in to the CCE console.
- Access the cluster console, choose Nodes in the navigation pane, and click the Node Pools tab on the right.
- upgrade is displayed in the Node Pool column of the existing nodes in the node pool.
- Click update. In the dialog box that is displayed, confirm whether to reset the node immediately.
Batch Synchronization
- Log in to the CCE console.
- Click the cluster name and access the cluster console. Choose Nodes in the navigation pane and click the Node Pools tab on the right.
- Choose More > Synchronize in the Operation column of the target node pool.
- In the Batch Synchronization window, set the synchronization parameters.
- Synchronization Policy: Node Reset is supported.
- Max. Nodes for Batch Synchronize: Nodes will be unavailable during synchronization in Node Reset mode. Set this parameter properly to prevent pod scheduling failures caused by too many unavailable nodes in the cluster.
- In the node list, select the nodes to be synchronized and click OK to start the synchronization.