@superlinked/sie-sdk 0.6.9 → 0.6.11

package/dist/index.d.cts

@@ -360,17 +360,25 @@ interface CapacityInfo {
     workers: WorkerInfo[];
 }
 /**
- * Pool specification for creating resource pools.
+ * Pool specification returned by the gateway.
  */
 interface PoolSpec {
     /** Pool name (used in GPU param as "poolName/machineProfile") */
     name: string;
+    /** Helm/NATS queue namespace backing this logical pool */
+    queue_pool?: string;
+    /** Optional bundle constraint for this pool */
+    bundle?: string | null;
     /** Machine profile requirements for pool readiness, e.g., { l4: 2, "a100-40gb": 1 } */
     gpus?: Record<string, number>;
-    /** Optional maximum assigned workers per machine profile */
-    gpuCaps?: Record<string, number>;
     /** Optional maximum assigned workers per machine profile, as returned by the gateway */
     gpu_caps?: Record<string, number>;
+    /** Optional TTL for dynamic pool leases */
+    ttl_seconds?: number | null;
+    /** Minimum workers kept warm for this pool */
+    minimum_worker_count?: number;
+    /** Models pinned for this pool */
+    pinned_models?: string[];
 }
 /**
  * Pool status information.
@@ -396,10 +404,7 @@ interface PoolInfo {
     /** Pool name */
     name: string;
     /** Pool specification */
-    spec: {
-        gpus?: Record<string, number>;
-        gpu_caps?: Record<string, number>;
-    };
+    spec: PoolSpec;
     /** Pool status */
     status: PoolStatus;
 }
@@ -924,7 +929,7 @@ declare function toFloat32Array(arr: number[]): Float32Array;
  * ```typescript
  * const client = new SIEClient("http://gateway:8080");
  *
- * // Create a dedicated pool
+ * // Create a logical pool backed by the cluster's default worker queue
  * await client.createPool("eval-bench", { l4: 2 });
  *
  * // Use pool for requests
@@ -1228,12 +1233,15 @@ declare class SIEClient {
     /**
      * Create or update a resource pool for isolated capacity.
      *
-     * Pools provide dedicated worker capacity, isolated from other clients.
-     * Workers are assigned to pools and only serve requests from that pool.
+     * Pools provide logical capacity isolation. By default they draw from the
+     * cluster's `default` Helm/NATS queue; pass `queuePool` only when the cluster
+     * has a dedicated physical worker queue declared under
+     * `queueRouting.staticQueuePools` for this workload.
      *
      * @param name - Pool name (used in GPU param as "poolName/machineProfile")
      * @param gpus - Optional machine profile requirements for pool readiness, e.g., { "l4": 2, "l4-spot": 1 }
      * @param gpuCaps - Optional maximum assigned workers per machine profile
+     * @param queuePool - Optional Helm/NATS queue namespace backing this logical pool. Defaults to "default".
      *
      * @example
      * ```typescript
@@ -1247,7 +1255,7 @@ declare class SIEClient {
      * await client.deletePool("eval-bench");
      * ```
      */
-    createPool(name: string, gpus?: Record<string, number>, gpuCaps?: Record<string, number>): Promise<void>;
+    createPool(name: string, gpus?: Record<string, number>, gpuCaps?: Record<string, number>, queuePool?: string): Promise<void>;
     /**
      * Get information about a pool.
      *
@@ -1360,7 +1368,7 @@ declare class SIEClient {
     private detectEndpointType;
 }
 
-declare const SDK_VERSION = "0.6.9";
+declare const SDK_VERSION = "0.6.11";
 
 /**
  * Helpers for converting SIE encode results to plain JavaScript types.

package/dist/index.d.ts

@@ -360,17 +360,25 @@ interface CapacityInfo {
     workers: WorkerInfo[];
 }
 /**
- * Pool specification for creating resource pools.
+ * Pool specification returned by the gateway.
  */
 interface PoolSpec {
     /** Pool name (used in GPU param as "poolName/machineProfile") */
     name: string;
+    /** Helm/NATS queue namespace backing this logical pool */
+    queue_pool?: string;
+    /** Optional bundle constraint for this pool */
+    bundle?: string | null;
     /** Machine profile requirements for pool readiness, e.g., { l4: 2, "a100-40gb": 1 } */
     gpus?: Record<string, number>;
-    /** Optional maximum assigned workers per machine profile */
-    gpuCaps?: Record<string, number>;
     /** Optional maximum assigned workers per machine profile, as returned by the gateway */
     gpu_caps?: Record<string, number>;
+    /** Optional TTL for dynamic pool leases */
+    ttl_seconds?: number | null;
+    /** Minimum workers kept warm for this pool */
+    minimum_worker_count?: number;
+    /** Models pinned for this pool */
+    pinned_models?: string[];
 }
 /**
  * Pool status information.
@@ -396,10 +404,7 @@ interface PoolInfo {
     /** Pool name */
     name: string;
     /** Pool specification */
-    spec: {
-        gpus?: Record<string, number>;
-        gpu_caps?: Record<string, number>;
-    };
+    spec: PoolSpec;
     /** Pool status */
     status: PoolStatus;
 }
@@ -924,7 +929,7 @@ declare function toFloat32Array(arr: number[]): Float32Array;
  * ```typescript
  * const client = new SIEClient("http://gateway:8080");
  *
- * // Create a dedicated pool
+ * // Create a logical pool backed by the cluster's default worker queue
  * await client.createPool("eval-bench", { l4: 2 });
  *
  * // Use pool for requests
@@ -1228,12 +1233,15 @@ declare class SIEClient {
     /**
      * Create or update a resource pool for isolated capacity.
      *
-     * Pools provide dedicated worker capacity, isolated from other clients.
-     * Workers are assigned to pools and only serve requests from that pool.
+     * Pools provide logical capacity isolation. By default they draw from the
+     * cluster's `default` Helm/NATS queue; pass `queuePool` only when the cluster
+     * has a dedicated physical worker queue declared under
+     * `queueRouting.staticQueuePools` for this workload.
      *
      * @param name - Pool name (used in GPU param as "poolName/machineProfile")
      * @param gpus - Optional machine profile requirements for pool readiness, e.g., { "l4": 2, "l4-spot": 1 }
      * @param gpuCaps - Optional maximum assigned workers per machine profile
+     * @param queuePool - Optional Helm/NATS queue namespace backing this logical pool. Defaults to "default".
      *
      * @example
      * ```typescript
@@ -1247,7 +1255,7 @@ declare class SIEClient {
      * await client.deletePool("eval-bench");
      * ```
      */
-    createPool(name: string, gpus?: Record<string, number>, gpuCaps?: Record<string, number>): Promise<void>;
+    createPool(name: string, gpus?: Record<string, number>, gpuCaps?: Record<string, number>, queuePool?: string): Promise<void>;
     /**
      * Get information about a pool.
      *
@@ -1360,7 +1368,7 @@ declare class SIEClient {
     private detectEndpointType;
 }
 
-declare const SDK_VERSION = "0.6.9";
+declare const SDK_VERSION = "0.6.11";
 
 /**
  * Helpers for converting SIE encode results to plain JavaScript types.

package/dist/index.js

@@ -854,7 +854,7 @@ function extractDataPayload(block) {
 }
 
 // src/version.ts
-var SDK_VERSION = "0.6.9";
+var SDK_VERSION = "0.6.11";
 
 // src/client.ts
 function sleep2(ms) {
@@ -1631,12 +1631,15 @@ var SIEClient = class {
   /**
    * Create or update a resource pool for isolated capacity.
    *
-   * Pools provide dedicated worker capacity, isolated from other clients.
-   * Workers are assigned to pools and only serve requests from that pool.
+   * Pools provide logical capacity isolation. By default they draw from the
+   * cluster's `default` Helm/NATS queue; pass `queuePool` only when the cluster
+   * has a dedicated physical worker queue declared under
+   * `queueRouting.staticQueuePools` for this workload.
    *
    * @param name - Pool name (used in GPU param as "poolName/machineProfile")
    * @param gpus - Optional machine profile requirements for pool readiness, e.g., { "l4": 2, "l4-spot": 1 }
    * @param gpuCaps - Optional maximum assigned workers per machine profile
+   * @param queuePool - Optional Helm/NATS queue namespace backing this logical pool. Defaults to "default".
    *
    * @example
    * ```typescript
@@ -1650,7 +1653,7 @@ var SIEClient = class {
    * await client.deletePool("eval-bench");
    * ```
    */
-  async createPool(name, gpus, gpuCaps) {
+  async createPool(name, gpus, gpuCaps, queuePool) {
     const alreadyTracking = this.pools.has(name);
     const requestBody = {
       name
@@ -1661,6 +1664,9 @@ var SIEClient = class {
     if (gpuCaps) {
       requestBody.gpu_caps = gpuCaps;
     }
+    if (queuePool) {
+      requestBody.queue_pool = queuePool;
+    }
     const url = `${this.baseUrl}/v1/pools`;
     const headers = {
       "Content-Type": JSON_CONTENT_TYPE,