Container Service for Kubernetes:View the health status of node instant scaling

When a node pool runs out of a specific instance type in a zone, scaling requests fail silently. The node instant scaling add-on monitors inventory levels across all configured zones and writes real-time status to a ConfigMap in the kube-system namespace. Check this ConfigMap to catch low-stock conditions before they block scale-out.

Prerequisites

Before you begin, make sure you have:

• A node pool with auto scaling enabled and node instant scaling turned on. See Enable Node Instant Elasticity.

• Node instant scaling add-on version 0.2.1 or later. To upgrade, see Manage add-ons.

How it works

After you enable instant scaling for a node pool, the node instant scaling add-on:

1. Retrieves all instance types configured for the node pool.

2. Polls real-time inventory status for each instance type in each configured zone.

3. Writes the results to a ConfigMap named {node pool id}-inventory-status in the kube-system namespace.

When supply status changes for any instance type in a configured zone, the add-on also emits an InstanceInventoryStatusChanged event on the ACKNodePool object, in the format:

{InstanceType}/{Zone} inventory status changed from {OldInventoryStatus} to {NewInventoryStatus}

Check inventory health status

1. Log on to the ACK console. In the left navigation pane, click Clusters.

2. On the Clusters page, click the name of the cluster.

3. In the left-side navigation pane, choose Configurations > ConfigMaps.

4. In the Namespace drop-down list, select kube-system. Find the ConfigMap named {node pool id}-inventory-status.

5. View the data.status field for real-time inventory health.

Understand the ConfigMap fields

The data.status field contains a JSON object. The examples below show a healthy and a degraded inventory state side by side so you can identify anomalies quickly.

Example: healthy inventory

{
    "SupplySeqs": [
        {
            "InstanceType": "ecs.g6.8xlarge",
            "Zone": "cn-hongkong-c",
            "State": "WithStock"
        },
        {
            "InstanceType": "ecs.g6.8xlarge",
            "Zone": "cn-hongkong-b",
            "State": "WithStock"
        },
        {
            "InstanceType": "ecs.g5ne.4xlarge",
            "Zone": "cn-hongkong-c",
            "State": "WithStock"
        },
        {
            "InstanceType": "ecs.g5ne.4xlarge",
            "Zone": "cn-hongkong-b",
            "State": "WithStock"
        }
    ],
    "Conditions": [
        {
            "Type": "InventoryNoStock",
            "Status": "False",
            "Reason": "NoInterruption",
            "Message": "all instanceTypes are with stock",
            "LastTransitionTime": "2024-08-01T08:20:49Z"
        }
    ],
    "Phase": "Healthy",
    "LastTransitionTime": "2024-08-01T08:20:49Z"
}

Example: degraded inventory

{
    "SupplySeqs": [
        {
            "InstanceType": "ecs.g6.8xlarge",
            "Zone": "cn-hongkong-c",
            "State": "WithoutStock"
        },
        {
            "InstanceType": "ecs.g6.8xlarge",
            "Zone": "cn-hongkong-b",
            "State": "WithStock"
        },
        {
            "InstanceType": "ecs.g5ne.4xlarge",
            "Zone": "cn-hongkong-c",
            "State": "ClosedWithoutStock"
        },
        {
            "InstanceType": "ecs.g5ne.4xlarge",
            "Zone": "cn-hongkong-b",
            "State": "WithStock"
        }
    ],
    "Conditions": [
        {
            "Type": "InventoryNoStock",
            "Status": "True",
            "Reason": "PartialInterruption",
            "Message": "some instanceTypes are out of stock",
            "LastTransitionTime": "2024-08-01T09:15:00Z"
        }
    ],
    "Phase": "UnHealthy",
    "LastTransitionTime": "2024-08-01T09:15:00Z"
}

What's next

View pod and node details and scaling activity history on the node scaling dashboard. See Node scaling dashboard.