Compute
Configure pre-warmed Kubernetes node pools for general, AI, and MCP workloads. Pre-warming uses balloon pods to keep Karpenter nodes provisioned for low-latency workload scheduling.
All endpoints require authentication via X-API-Key header and the appropriate scope.
Pre-Warmed Config Object
{
"general": {
"enabled": true,
"count": 2,
"instanceCategory": "m",
"instanceSize": "large"
},
"ai": {
"enabled": true,
"count": 1,
"instanceCategory": "g",
"instanceSize": "xlarge"
},
"mcp": {
"enabled": false,
"count": 0,
"instanceCategory": "m",
"instanceSize": "medium"
}
}
GET /api/v1/compute/pre-warmed
Get pre-warmed node configurations for all workload types.
Scope: compute:read
Response 200 OK
Returns the full Pre-Warmed Config object.
PUT /api/v1/compute/pre-warmed/:workloadType
Configure pre-warmed nodes for a workload type. Creates balloon pods to keep Karpenter nodes provisioned.
Scope: compute:write
Path Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
workloadType | string | Yes | Workload type: general, ai, or mcp |
Request Body
{
"enabled": true,
"count": 2,
"instanceCategory": "m",
"instanceSize": "large"
}
| Field | Type | Required | Description |
|---|---|---|---|
enabled | boolean | Yes | Enable/disable pre-warming |
count | number | Yes | Number of pre-warmed nodes (1-5) |
instanceCategory | string | Yes | Instance category (m, c, g, etc.) |
instanceSize | string | Yes | Instance size (small, medium, large, xlarge, 2xlarge, 4xlarge) |
Response 200 OK
{
"workloadType": "general",
"enabled": true,
"count": 2,
"instanceCategory": "m",
"instanceSize": "large"
}
DELETE /api/v1/compute/pre-warmed/:workloadType
Disable pre-warmed nodes for a workload type and remove balloon pods.
Scope: compute:write
Path Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
workloadType | string | Yes | Workload type: general, ai, or mcp |
Response 200 OK
{
"workloadType": "general",
"disabled": true
}