fix(slurm/slinky):

charts/topograph/values.slinky.block-example.yaml

@@ -19,7 +19,7 @@ global:
         matchLabels:
           app.kubernetes.io/component: compute
       plugin: topology/block
-      block_sizes: 4
+      blockSizes: [4]
       topologyConfigPath: topology.conf
       topologyConfigmapName: slurm-config
       reportingMode: staticNodes

docs/api.md

@@ -85,13 +85,13 @@ Topograph exposes three endpoints for interacting with the service. Below are th
     - **slurm parameters**:
       - **topologyConfigPath**: (optional) A string specifying the file path for the topology configuration. If omitted, the topology config content is returned in the HTTP response.
       - **plugin**: (optional) A string specifying topology plugin: `topology/tree` (default) or `topology/block`.
-      - **block_sizes**: (optional) A string specifying block size for `topology/block` plugin.
+      - **blockSizes**: (optional) A string specifying block size for `topology/block` plugin.
       - **reconfigure**: (optional) If `true`, invoke `scontrol reconfigure` after topology config is generated. Default `false`
     - **slinky parameters**:
       - **namespace**: A string specifying namespace where SLURM cluster is running.
       - **podSelector**: A standard Kubernetes label selector for pods running SLURM nodes.
       - **plugin**: (optional) A string specifying topology plugin: `topology/tree` (default) or `topology/block`.
-      - **block_sizes**: (optional) A string specifying block size for `topology/block` plugin.
+      - **blockSizes**: (optional) A string specifying block size for `topology/block` plugin.
       - **topologyConfigPath**: A string specifying the key for the topology config in the ConfigMap.
       - **topologyConfigmapName**: A string specifying the name of the ConfigMap containing the topology config.
   - **nodes**: (optional) An array of regions mapping instance IDs to node names.
@@ -111,7 +111,7 @@ Topograph exposes three endpoints for interacting with the service. Below are th
     "name": "slurm",
     "params": {
       "plugin": "topology/block",
-      "block_sizes": "30,120"
+      "blockSizes": [30,120]
     }
   },
   "nodes": [

docs/engines/slinky.md

@@ -31,14 +31,14 @@ global:
   provider: aws
   engine: slinky
   engineParams:
-    namespace: ns-slinky                               # Namespace where Slinky is running
-    podSelector:                                      # Label selector for pods running SLURM nodes
+    namespace: ns-slinky                       # Namespace where Slinky is running
+    podSelector:                               # Label selector for pods running SLURM nodes
       matchLabels:
         app.kubernetes.io/component: compute
-    plugin: topology/block                             # Name of the topology plugin
-    block_sizes: 4                                     # (Optional) Block size for the block topology plugin
-    topologyConfigmapName: slurm-config              # Name of the ConfigMap containing the topology config
-    topologyConfigPath: topology.conf                # Key in the ConfigMap for the topology config
+    plugin: topology/block                     # Name of the topology plugin
+    blockSizes: [4]                            # (Optional) Block size for the block topology plugin
+    topologyConfigmapName: slurm-config        # Name of the ConfigMap containing the topology config
+    topologyConfigPath: topology.conf          # Key in the ConfigMap for the topology config
 ```
 
 ### Per-partition topologies
@@ -160,7 +160,7 @@ curl -X POST -H "Content-Type: application/json" \
         "topologyConfigPath": "topology.conf",
         "topologyConfigmapName": "slurm-config",
         "plugin": "topology/block",
-        "block_sizes": "8,16,32"
+        "blockSizes": [8,16,32]
       }
     }
   }' \

docs/engines/slurm.md

@@ -58,7 +58,7 @@ id=$(curl -s -X POST -H "Content-Type: application/json" -d '{"provider":{"param
 
 And if you want to view the block topology (with specified block sizes), use the command:
 ```bash
-id=$(curl -s -X POST -H "Content-Type: application/json" -d '{"provider":{"params":{"modelFileName":"/usr/local/bin/tests/models/<cluster-model>.yaml"}},"engine":{"params":{"plugin":"topology/block", "block_sizes": "4,8"}}}' http://localhost:49021/v1/generate)
+id=$(curl -s -X POST -H "Content-Type: application/json" -d '{"provider":{"params":{"modelFileName":"/usr/local/bin/tests/models/<cluster-model>.yaml"}},"engine":{"params":{"plugin":"topology/block", "blockSizes": [4,8]}}}' http://localhost:49021/v1/generate)
 ```
 
 You can query the results of either topology request with:

docs/get-started/quickstart-k8s.md

@@ -76,7 +76,7 @@ Engine-specific install flags point the `slinky` engine at the Slinky deployment
   --set global.engineParams.topologyConfigPath=topology.conf
 ```
 
-The full engine-parameter shape (`podSelector`, `plugin`, `block_sizes`, per-partition topologies, …) is documented in the [chart README](https://github.com/NVIDIA/topograph/blob/main/charts/topograph/README.md). Example values files for common Slinky scenarios ship in the chart directory:
+The full engine-parameter shape (`podSelector`, `plugin`, `blockSizes`, per-partition topologies, …) is documented in the [chart README](https://github.com/NVIDIA/topograph/blob/main/charts/topograph/README.md). Example values files for common Slinky scenarios ship in the chart directory:
 
 - [`values.slinky.tree-example.yaml`](https://github.com/NVIDIA/topograph/blob/main/charts/topograph/values.slinky.tree-example.yaml) — tree topology
 - [`values.slinky.block-example.yaml`](https://github.com/NVIDIA/topograph/blob/main/charts/topograph/values.slinky.block-example.yaml) — block topology

pkg/engines/slinky/engine.go

@@ -22,6 +22,8 @@ import (
 	"fmt"
 	"maps"
 	"net/http"
+	"strconv"
+	"strings"
 	"time"
 
 	corev1 "k8s.io/api/core/v1"
@@ -258,7 +260,7 @@ func (eng *SlinkyEngine) generateConfigMapAnnotations() map[string]string {
 		annotations[topology.KeyConfigMapPlugin] = eng.params.Plugin
 	}
 	if len(eng.params.BlockSizes) != 0 {
-		annotations[topology.KeyConfigMapBlockSizes] = eng.params.BlockSizes
+		annotations[topology.KeyConfigMapBlockSizes] = intToStr(eng.params.BlockSizes)
 	}
 
 	return annotations
@@ -545,3 +547,11 @@ func (eng *SlinkyEngine) updateNodeAnnotation(ctx context.Context, node *corev1.
 
 	return nil
 }
+
+func intToStr(input []int) string {
+	strs := make([]string, len(input))
+	for i, n := range input {
+		strs[i] = strconv.Itoa(n)
+	}
+	return strings.Join(strs, ",")
+}

pkg/engines/slinky/engine_test.go

@@ -143,14 +143,14 @@ func TestGetParameters(t *testing.T) {
 				topology.KeyPodSelector:       podSelector,
 				topology.KeyNodeSelector:      nodeSelector,
 				topology.KeyPlugin:            topology.TopologyBlock,
-				topology.KeyBlockSizes:        "16",
+				topology.KeyBlockSizes:        []int{16},
 				topology.KeyTopoConfigPath:    "path",
 				topology.KeyTopoConfigmapName: "name",
 			},
 			ret: &Params{
 				BaseParams: slurm.BaseParams{
 					Plugin:     topology.TopologyBlock,
-					BlockSizes: "16",
+					BlockSizes: []int{16},
 				},
 				Namespace:     "namespace",
 				PodSelector:   labelSelector,
@@ -343,7 +343,7 @@ func TestConfigMapAnnotationsAndMetadata(t *testing.T) {
 			name: "with block sizes only",
 			params: &Params{
 				BaseParams: slurm.BaseParams{
-					BlockSizes: "8,16,32",
+					BlockSizes: []int{8, 16, 32},
 				},
 				Namespace:     "test-namespace",
 				PodSelector:   labelSelector,
@@ -357,7 +357,7 @@ func TestConfigMapAnnotationsAndMetadata(t *testing.T) {
 			params: &Params{
 				BaseParams: slurm.BaseParams{
 					Plugin:     topology.TopologyBlock,
-					BlockSizes: "8,16,32",
+					BlockSizes: []int{8, 16, 32},
 				},
 				Namespace:     "test-namespace",
 				PodSelector:   labelSelector,
@@ -388,19 +388,19 @@ func TestConfigMapAnnotationsAndMetadata(t *testing.T) {
 				require.NotContains(t, annotations, topology.KeyConfigMapPlugin)
 			}
 			if tc.wantBlock {
-				requireAnnotation(t, annotations, topology.KeyConfigMapBlockSizes, tc.params.BlockSizes)
+				requireAnnotation(t, annotations, topology.KeyConfigMapBlockSizes, intToStr(tc.params.BlockSizes))
 			} else {
 				require.NotContains(t, annotations, topology.KeyConfigMapBlockSizes)
 			}
 
 			// Metadata logic (simulate GenerateOutput)
 			tree := &topology.Vertex{Name: "root"}
-			setMetadata(tree, tc.params.Plugin, tc.params.BlockSizes)
+			setMetadata(tree, tc.params.Plugin, intToStr(tc.params.BlockSizes))
 			if tc.wantPlugin {
 				require.Equal(t, tc.params.Plugin, tree.Metadata[topology.KeyPlugin])
 			}
 			if tc.wantBlock {
-				require.Equal(t, tc.params.BlockSizes, tree.Metadata[topology.KeyBlockSizes])
+				require.Equal(t, intToStr(tc.params.BlockSizes), tree.Metadata[topology.KeyBlockSizes])
 			}
 		})
 	}

pkg/engines/slurm/slurm.go

@@ -23,7 +23,6 @@ import (
 	"fmt"
 	"net/http"
 	"regexp"
-	"strconv"
 	"strings"
 
 	"k8s.io/klog/v2"
@@ -34,7 +33,6 @@ import (
 	"github.com/NVIDIA/topograph/internal/files"
 	"github.com/NVIDIA/topograph/internal/httperr"
 	"github.com/NVIDIA/topograph/pkg/engines"
-	"github.com/NVIDIA/topograph/pkg/metrics"
 	"github.com/NVIDIA/topograph/pkg/topology"
 	"github.com/NVIDIA/topograph/pkg/translate"
 )
@@ -51,10 +49,8 @@ const NAME = "slurm"
 type SlurmEngine struct{}
 
 type BaseParams struct {
-	Plugin           string `mapstructure:"plugin"`
-	BlockSizes       string `mapstructure:"block_sizes"`
-	FakeNodesEnabled bool   `mapstructure:"fakeNodesEnabled"`
-	FakeNodePool     string `mapstructure:"fake_node_pool"`
+	Plugin     string `mapstructure:"plugin"`
+	BlockSizes []int  `mapstructure:"blockSizes"`
 }
 
 type Topology struct {
@@ -184,34 +180,6 @@ func GetNodeList(ctx context.Context) ([]string, error) {
 	return nodes, nil
 }
 
-func GetFakeNodes(ctx context.Context) (string, error) {
-	args := []string{"show", "partition", "fake"}
-	stdout, err := exec.Exec(ctx, "scontrol", args, nil)
-	if err != nil {
-		return "", err
-	}
-	out := stdout.String()
-	klog.V(4).Infof("stdout: %s", out)
-
-	return parseFakeNodes(out)
-}
-
-func parseFakeNodes(data string) (string, error) {
-	prefix := "Nodes="
-	scanner := bufio.NewScanner(strings.NewReader(data))
-	for scanner.Scan() {
-		if line := strings.TrimSpace(scanner.Text()); strings.HasPrefix(line, prefix) {
-			return line[len(prefix):], nil
-		}
-	}
-
-	if err := scanner.Err(); err != nil {
-		return "", fmt.Errorf("failed to scan fake nodes partition: %v", err)
-	}
-
-	return "", fmt.Errorf("fake partition has no nodes")
-}
-
 func getPartitionNodes(ctx context.Context, partition string, _ []any) (string, error) {
 	args := []string{"show", "partition", partition}
 	stdout, err := exec.Exec(ctx, "scontrol", args, nil)
@@ -307,22 +275,7 @@ func GenerateOutputParams(ctx context.Context, root *topology.Vertex, params *Pa
 func GetTranslateConfig(ctx context.Context, params *BaseParams, topologies map[string]*Topology, f *TopologyNodeFinder) (*translate.Config, error) {
 	cfg := &translate.Config{
 		Plugin:     params.Plugin,
-		BlockSizes: getBlockSizes(params.BlockSizes),
-	}
-
-	// set fake nodes
-	if params.Plugin == topology.TopologyBlock && params.FakeNodesEnabled {
-		var fakeNodes string
-		var err error
-		if len(params.FakeNodePool) > 0 {
-			fakeNodes = params.FakeNodePool
-		} else {
-			fakeNodes, err = GetFakeNodes(ctx)
-			if err != nil {
-				return nil, err
-			}
-		}
-		cfg.FakeNodePool = fakeNodes
+		BlockSizes: params.BlockSizes,
 	}
 
 	// set per-partition topologies
@@ -361,24 +314,6 @@ func getParams(params map[string]any) (*Params, error) {
 	return &p, err
 }
 
-func getBlockSizes(str string) []int {
-	if len(str) == 0 {
-		return nil
-	}
-	parts := strings.Split(str, ",")
-	blockSizes := make([]int, 0, len(parts))
-	for _, part := range parts {
-		sz, err := strconv.Atoi(part)
-		if err != nil {
-			metrics.AddValidationError("BlockSize parsing error")
-			klog.Warningf("Failed to parse blockSize %v: %v. Ignoring admin blockSizes.", part, err)
-			return nil
-		}
-		blockSizes = append(blockSizes, sz)
-	}
-	return blockSizes
-}
-
 func reconfigure(ctx context.Context) error {
 	stdout, err := exec.Exec(ctx, "scontrol", []string{"reconfigure"}, nil)
 	if err != nil {