Node pool scale-out can fail when the selected zones run out of instance inventory or the chosen instance types are unavailable in a zone. ACK's elastic strength metric measures how many resource pool combinations in your node pool can actually provision new nodes. ACK flags node pools with a poor score so you can identify and fix configuration issues before — or after — a scale-out failure.
How it works
Not every theoretical resource pool can provision nodes at any given time. Inventory levels, resource constraints, and configuration mismatches can reduce the number of available resource pools below the theoretical maximum. For example, the c6e instance family supports enterprise SSDs but not standard SSDs.
The following figure shows a node pool configured with two zones, two instance types, and a specific disk type. Although multiple resource pools exist in theory, only one is actually available due to inventory shortages, instance deployment status, and disk resource constraints.
View scalability details
Log on to the Container Service Management Console. In the left navigation pane, click Clusters.
On the Clusters page, click the name of the target cluster. In the left navigation pane, choose Nodes > Node Pools.
On the Node Pools page, check the Status column. A Scalability Level: Weak label indicates that the node pool's elastic strength score is 0 or less.
Click Scalability Level: Weak to open the Scalability Details dialog box and view the assessment results.
Understand and fix scalability issues
The Scalability Details dialog box shows which resource pools are unavailable and why. Issues fall into two categories.
Resource constraint issues
Configuration mismatches that prevent a resource pool from being valid. Fix these by adjusting your node pool configuration.
| Result | Recommended fix |
|---|---|
| Image and instance type mismatch | Adjust the instance types in the node pool |
| System disk and instance type mismatch | Adjust the instance types or system disk type in the node pool |
| Data disk and instance type mismatch | Adjust the instance types or data disk type in the node pool |
| Instance type does not support IPv6 | Adjust the instance types in the node pool |
| vSwitch does not exist | Adjust the vSwitches in the node pool |
Instance inventory issues
The instance type exists and is valid in the zone, but is temporarily unavailable. Expand to other zones or instance types to increase the number of available resource pools.
| Result | Recommended fix |
|---|---|
| Instance type not deployed | Adjust the instance types in the node pool |
| Instance type is out of stock | Adjust the instance types in the node pool |
Improving elastic strength
Each valid zone–instance type combination adds one more resource pool, raising your elastic strength score. To increase available resource pools and improve high availability, configure multiple vSwitches across different zones and specify multiple instance types with sufficient inventory.
To adjust node pool configuration, see Create and manage a node pool.
What's next
ECS instance type configuration recommendations — choose instance types that are widely available across zones to maximize resource pool count.