chore(topograph): Updated the recipes to use topograph v1.0.0 (#2254)

Signed-off-by: Ravi Shankar <ravish@nvidia.com>

Author: ravisoundar

docs/integrator/recipe-development.md

@@ -171,11 +171,10 @@ Leaves that need topology-aware scheduling can optionally add `slinky-topograph`
     - slinky-slurm-operator-crds
     - slinky-slurm   # deploy after Slurm so Helm already owns config-extra
   overrides:
-    global:
-      provider:
-        name: gcp      # cloud provider: gcp | aws | oci | nebius | …
-      engine:
-        name: slinky   # scheduler consumer: slinky | slurm | k8s | graph
+    provider:
+      name: gcp      # cloud provider: gcp | aws | oci | nebius | …
+    engine:
+      name: slinky   # scheduler consumer: slinky | slurm | k8s | graph
 
 - name: slinky-slurm
   type: Helm
@@ -205,9 +204,8 @@ For cloud providers (gcp, aws, oci, nebius, …), `slinky-topograph` requires IA
 
 ```yaml
 overrides:
-  global:
-    provider:
-      name: gcp
+  provider:
+    name: gcp
   serviceAccount:
     annotations:
       iam.gke.io/gcp-service-account: <sa-name>@<project-id>.iam.gserviceaccount.com

docs/user/component-catalog.md

@@ -43,7 +43,7 @@ The source of truth is [`recipes/registry.yaml`](https://github.com/NVIDIA/aicr/
 | **slinky-slurm-operator-crds** | Custom Resource Definitions for the SchedMD Slinky Slurm operator. Installs the `slinky.slurm.net` CRDs (Controller, NodeSet, LoginSet, Accounting, RestApi, Token). Installed separately to support CRD lifecycle management. | [Slinky Slurm Operator](https://github.com/SlinkyProject/slurm-operator) |
 | **slinky-slurm-operator** | SchedMD Slinky Slurm operator and admission webhook. Manages the lifecycle of Slurm clusters declared via Slinky CRs (Controller, NodeSet, LoginSet, Accounting, RestApi, Token). AICR's system node-selector and toleration bundle flags apply to both deployments; affinity remains available through component values or typed overrides. | [Slinky Slurm Operator](https://github.com/SlinkyProject/slurm-operator) |
 | **slinky-slurm** | Slinky-managed Slurm cluster instance: Controller (slurmctld) + LoginSet (sackd/sshd) + NodeSet (slurmd) + RestApi (slurmrestd), with SlurmDBD derived from the recipe's typed accounting mode. Reconciled by `slinky-slurm-operator`. See [Slurm Accounting](slinky-slurm-accounting.md), [Slurm Enroot Configuration](slinky-slurm-enroot.md), and [Slurm Shared Storage](slinky-slurm-storage.md). | [Slinky Slurm Cluster Chart](https://github.com/SlinkyProject/slurm-operator/tree/main/helm/slurm) |
-| **slinky-topograph** | Slinky/Slurm-scoped instance of Topograph — queries cloud provider topology APIs (GCP, AWS, OCI …) to generate Slurm `topology.conf`, enabling topology-aware placement decisions in the Slinky-managed scheduler. **Not installed by default**; leaf overlays opt in by adding an explicit `componentRef` entry for `slinky-topograph` — the `componentRef` is what schedules the release; `dependencyRefs` alone does not install anything. That `componentRef` declares `slinky-slurm` as a `dependencyRef` to deploy **after** it: `slinky-slurm` renders and owns the `slinky-slurm-config-extra` ConfigMap (from its `configFiles`, mounted into slurmctld via the Controller CR's `configFileRefs`), and Topograph patches only that ConfigMap's `topology.conf` key on each sync, preserving the chart-owned `cgroup.conf`/`gres.conf` keys — Helm has to own the ConfigMap first. `TopologyPlugin: topology/tree` is set per-leaf via `slinky-slurm`'s `controller.extraConfMap`. Includes the `node-observer` sub-chart, which watches the topograph API pod and regenerates topology on restarts or selected node/pod changes. Requires cloud provider IAM access (e.g. GCP `roles/compute.viewer` for Workload Identity). | [Topograph](https://github.com/NVIDIA/topograph) |
+| **slinky-topograph** | Slinky/Slurm-scoped instance of Topograph — queries cloud provider topology APIs (GCP, AWS, OCI …) to generate Slurm `topology.conf`, enabling topology-aware placement decisions in the Slinky-managed scheduler. **Not installed by default**; leaf overlays opt in by adding an explicit `componentRef` entry for `slinky-topograph` — the `componentRef` is what schedules the release; `dependencyRefs` alone does not install anything. That `componentRef` declares `slinky-slurm` as a `dependencyRef` to deploy **after** it: `slinky-slurm` renders and owns the `slinky-slurm-config-extra` ConfigMap (from its `configFiles`, mounted into slurmctld via the Controller CR's `configFileRefs`), and Topograph patches only that ConfigMap's `topology.conf` key on each sync, preserving the chart-owned `cgroup.conf`/`gres.conf` keys — Helm has to own the ConfigMap first. `TopologyPlugin: topology/tree` is set per-leaf via `slinky-slurm`'s `controller.extraConfMap`. Includes the `node-observer` component, which watches the topograph API pod and regenerates topology on restarts or selected node/pod changes. Requires cloud provider IAM access (e.g. GCP `roles/compute.viewer` for Workload Identity). | [Topograph](https://github.com/NVIDIA/topograph) |
 | **nfd-ocp-olm** | OLM installer for Node Feature Discovery on OpenShift. Creates the OperatorGroup and Subscription resources that install NFD via the Operator Lifecycle Manager. Paired with `nfd-ocp`. OCP-specific. | [Node Feature Discovery (Certified)](https://catalog.redhat.com/software/container-stacks/detail/5ec53e8c110f56bd24f5f8db) |
 | **nfd-ocp** | Node Feature Discovery CR for OpenShift. Configures NFD's operand (worker, topology updater) via a NodeFeatureDiscovery custom resource. Deployed after `nfd-ocp-olm`. OCP-specific. | [Node Feature Discovery](https://github.com/kubernetes-sigs/node-feature-discovery) |
 | **gpu-operator-ocp-olm** | OLM installer for the GPU Operator on OpenShift. Creates the OperatorGroup and Subscription resources that install the certified GPU Operator via the Operator Lifecycle Manager. Paired with `gpu-operator-ocp`. OCP-specific. | [NVIDIA GPU Operator (Certified)](https://catalog.redhat.com/software/container-stacks/detail/5e7b210b8a3c1e00013d636d) |

docs/user/container-images.md

@@ -72,7 +72,7 @@ _Rendering fidelity:_ `catalog-parity: charts are rendered with the shared recip
 | slinky-slurm | helm | slurm | 1.2.0 | 5 |
 | slinky-slurm-operator | helm | slurm-operator | 1.2.0 | 2 |
 | slinky-slurm-operator-crds | helm | slurm-operator-crds | 1.2.0 | 0 |
-| slinky-topograph | helm | topograph/topograph | 0.5.0 | 1 |
+| slinky-topograph | helm | topograph/topograph | 1.0.0 | 1 |
 | slurm-accounting-mariadb | helm | mariadb-cluster | 26.6.0 | 0 |
 
 ## Version variants
@@ -320,7 +320,7 @@ _No images extracted._
 
 ### slinky-topograph
 
-- `ghcr.io/nvidia/topograph:v0.5.0`
+- `ghcr.io/nvidia/topograph:v1.0.0`
 
 ### slurm-accounting-mariadb
 

pkg/bundler/testdata/stock_render_golden.yaml

@@ -29,10 +29,10 @@ h100-eks-ubuntu-training-kubeflow: 8c1fe9721baba50d2d383e30d883c92f8c2cb99a881ed
 h100-eks-ubuntu-training-slurm: 082a43adc19c769b31313e09c685c18b779321b2dbcbc67a8f5699dd5ae61826
 h100-gke-cos-inference-dynamo: a06816f00f95b6d8dd134b38ae0d2d8c5c36ab6d643181c4c9651ac409e7b5a5
 h100-gke-cos-training-kubeflow: ac2c6cf43293c88d04528382cb5ba189ff906489b2c95ce636a6d0f29be5f43f
-h100-gke-cos-training-slurm: 9029621222d4c5f5cd49bc380e0aaa9ea6dadf7debbac45fd301ece293c29bd8
+h100-gke-cos-training-slurm: 8da5bafdf36f1c9118f75f32b1e0fad1d1b101daadb89036c4564c6b91dc4b00
 h100-kind-inference-dynamo: 32d919b1af2f780a989dd044203cb7220593b081c822227b91810249f4cad821
 h100-kind-training-kubeflow: dd2ded798dbc2463fafd3da93126dc431f2e42410e060e7cc4e0be4cd4d39fa9
-h100-kind-training-slurm: 96d9710ffbf51c1cd62d2434c816601aff8969f44dc02fba3f645fbc78b0dbd9
+h100-kind-training-slurm: 1b7147dd701c02e3db15032fbe7f34de87839bea71f0cfbaa5f5463e599e5b3f
 h200-any: b9e3f4593b7cced8ab233b44d52488a6fac5eeac3f9ca0a7cb1ee214672f72a1
 h200-eks-inference: b6454b224e2287eec1f8608bf539679cf8a7fbc7016bebdc139afc944315a414
 h200-eks-training: 9c0c79822ed70e2010158af6eba120ee0d8f942f1fafd9d981f342394cae0929

pkg/recipe/testdata/catalog_parity_golden.yaml

@@ -29,10 +29,10 @@ h100-eks-ubuntu-training-kubeflow: d0dc8163f513470f8cdcb7906d2ff6eef630185c3469a
 h100-eks-ubuntu-training-slurm: 04fb61932ca46d4b0ae9f232d0c40dbc9b02872430091849c44dcf43356f366f
 h100-gke-cos-inference-dynamo: b5eb7a000dd238471584b7daf062f1867da4cd3c2468343a1f79582f6ec1f38a
 h100-gke-cos-training-kubeflow: ed11489b48c4a036b4e3d336ec41b1776e286f15a8e0caa75315e623074e4385
-h100-gke-cos-training-slurm: 32bb46642f1f234796d4e4ed0791e04f7d8c2a3ddb95312f8051a34a052b1988
+h100-gke-cos-training-slurm: 43cc50555b76ddadf175a45df49fe4f50713c1df2a40afd74a9a421eca8ba8bd
 h100-kind-inference-dynamo: f665ecad511b4492eb79f6de19d4ccb9e2721fcdf34ead85711b00b52678e1f0
 h100-kind-training-kubeflow: da0ba5fd7872854337cfca957880cbe124317c154b61608a8b21c55d2172d6d8
-h100-kind-training-slurm: 678a468fd348bbd2922cf5a282abef060b285410699bbca16cfbd5a7a890dc8b
+h100-kind-training-slurm: 184c5240e6956fe0f2e3310922044ecdfbb8cb4f0df815d696b561cb49e60bf1
 h200-any: b6bdb82e355993c99840147c3846656aff004a1fb4253fe2d48b85bfd860ce4d
 h200-eks-inference: c3c801df830173e967a366f31462ace88079656a0a660deafa09d7d45ee91583
 h200-eks-training: 3cb65fd9e45d7f6fbf440469c7a27ad975b634850f6ebd2566831c54dfa7a2ba

recipes/components/slinky-topograph/manifests/kind-test-model.yaml

@@ -22,44 +22,45 @@
 # it deploys after slinky-slurm so that chart owns the config-extra
 # ConfigMap before Topograph patches it.
 #
-# global.provider.name is the REQUIRED per-overlay override — set it in the
-# consuming overlay's overrides: block. global.engine.name is pinned to
-# slinky below: the chart schema requires global.engine, but this shared
-# values file already satisfies it, so overlays do not need to repeat it
-# (the slurm leaves still set it explicitly, which is harmless).
-#   Valid providers (chart values.schema.json @ v0.5.0): aws, aws-sim, cw,
-#   dra, dsx-sim, gcp, gcp-sim, infiniband-bm, infiniband-k8s, lambdai,
+# provider.name is the REQUIRED per-overlay override — set it in the
+# consuming overlay's overrides: block. engine.name already defaults to
+# slinky below: the chart schema requires engine, but this shared values
+# file already satisfies it, so overlays do not need to repeat it (the
+# slurm leaves still set it explicitly, which is harmless).
+#   Valid providers (chart values.schema.json @ v1.0.0): aws, aws-sim, dra,
+#   dsx-sim, gcp, gcp-sim, infiniband-bm, infiniband-k8s, lambdai,
 #   lambdai-sim, nebius, nebius-sim, netq, nscale, nscale-sim, oci,
 #   oci-imds, oci-sim, test
-#   Valid engines: graph, k8s, slinky, slurm
+#   Valid engines: graph, k8s, nfd, slinky, slurm
 # Example (GKE + Slinky):
 #   overrides:
-#     global:
-#       provider:
-#         name: gcp
+#     provider:
+#       name: gcp
 
 # Stabilize the Deployment name to "slinky-topograph"; health-check assertions
-# and dependencyRef consumers rely on it.
+# and dependencyRef consumers rely on it. nodeObserver and nodeDataBroker
+# Deployment/DaemonSet names derive from this fullname (no separate
+# fullnameOverride for those components — they are rendered from the same
+# chart, not standalone sub-charts).
 fullnameOverride: slinky-topograph
 
-global:
-  service:
-    type: ClusterIP
-    port: 49021
-  engine:
-    name: slinky
-    params:
-      plugin: topology/tree
-      podSelector:
-        matchLabels:
-          app.kubernetes.io/name: slurmd
-      # Chart-owned ConfigMap rendered by the slurm chart from .Values.configFiles
-      # (<fullname>-config-extra) and mounted into slurmctld via the Controller
-      # CR's configFileRefs. Topograph patches ONLY the topology.conf key
-      # (create-or-update with key-level merge), preserving cgroup.conf/gres.conf.
-      topologyConfigmapName: slinky-slurm-config-extra
-      topologyConfigPath: topology.conf
-      namespace: slurm
+service:
+  type: ClusterIP
+  port: 49021
+engine:
+  name: slinky
+  params:
+    plugin: topology/tree
+    podSelector:
+      matchLabels:
+        app.kubernetes.io/name: slurmd
+    # Chart-owned ConfigMap rendered by the slurm chart from .Values.configFiles
+    # (<fullname>-config-extra) and mounted into slurmctld via the Controller
+    # CR's configFileRefs. Topograph patches ONLY the topology.conf key
+    # (create-or-update with key-level merge), preserving cgroup.conf/gres.conf.
+    topologyConfigmapName: slinky-slurm-config-extra
+    topologyConfigPath: topology.conf
+    namespace: slurm
 
 # Topology request batching delay. Topograph waits this long after the first
 # request before querying the provider API, collecting parallel requests into
@@ -72,19 +73,17 @@ serviceAccount:
   create: true
   name: slinky-topograph
 
-# node-observer sub-chart — watches the Topograph API pod and triggers
-# topology regeneration on API restarts or selected node/pod changes.
-# Bundled as a local chart dependency; always deployed alongside topograph.
-node-observer:
-  # Stabilize the Deployment name for health-check assertions.
-  fullnameOverride: slinky-topograph-node-observer
+# Node Observer — watches the Topograph API pod and triggers topology
+# regeneration on API restarts or selected node/pod changes. Rendered from
+# the same chart as topograph (not a sub-chart); always deployed alongside it.
+nodeObserver:
   topograph:
     trigger:
       podSelector:
         matchLabels:
           app.kubernetes.io/name: slurmd
 
-# node-data-broker sub-chart
-# Bundled as a local chart dependency.
-node-data-broker:
-  fullnameOverride: slinky-topograph-node-data-broker
+# Node Data Broker — rendered from the same chart as topograph (not a
+# sub-chart). Enabled by default; leaves that don't need it set
+# nodeDataBroker.enabled: false in their overlay overrides.
+nodeDataBroker: {}

recipes/components/slinky-topograph/values.yaml

@@ -102,11 +102,10 @@ spec:
         - slinky-slurm-operator-crds
         - slinky-slurm
       overrides:
-        global:
-          provider:
-            name: gcp
-          engine:
-            name: slinky
+        provider:
+          name: gcp
+        engine:
+          name: slinky
 
     - name: slinky-slurm
       type: Helm

recipes/overlays/h100-gke-cos-training-slurm.yaml

@@ -82,36 +82,36 @@ spec:
         - slinky-slurm-operator
         - slinky-slurm-operator-crds
         - slinky-slurm
-      preManifestFiles:
-        - components/slinky-topograph/manifests/kind-test-model.yaml
       overrides:
-        global:
-          provider:
-            name: test
-            params:
-              modelFileName: /etc/topograph/models/kind-slinky.yaml
-          engine:
-            name: slinky
-        # Disable the node-data-broker in the kind environment.
-        node-data-broker:
+        provider:
+          name: test
+          params:
+            # Bare filename only: the v1.0.0 test provider's model loader
+            # resolves this against topograph's embedded tests/models/*.yaml
+            # (compiled into the binary) and rejects any path component —
+            # a mounted ConfigMap file cannot be referenced this way. Use one
+            # of topograph's built-in fixtures (tests/models/small-tree.yaml
+            # upstream) instead of a custom mounted model.
+            modelFileName: small-tree.yaml
+        engine:
+          name: slinky
+        # Disable the node-data-broker in the kind environment. nodeObserver
+        # automatically skips waiting on it when nodeDataBroker.enabled is false.
+        nodeDataBroker:
           enabled: false
-        node-observer:
-          topograph:
-            nodeDataBroker:
-              enabled: false
-        volumes:
-          - name: test-model
-            configMap:
-              name: slinky-topograph-test-model
-        volumeMounts:
-          - name: test-model
-            mountPath: /etc/topograph/models
-            readOnly: true
       # Inline override of the registry assertFile (checks/slinky-topograph/
       # health-check.yaml): this leaf disables node-data-broker, so the
-      # DaemonSet step must not run, and the test provider's mounted
-      # kind-slinky.yaml model (references the real slinky-0 node) makes
-      # the exact topology.conf content assertable.
+      # DaemonSet step must not run, and the exact topology.conf content
+      # below pins the small-tree.yaml fixture's own shape — it does NOT
+      # validate resolution of the leaf's real one-replica worker
+      # (slinky-0). Topograph's test provider always returns its fixture
+      # data as a ready-made instance map, and when the engine already has
+      # a non-empty instance map it uses that as-is instead of resolving
+      # real pod/node names — so the real slinky-0 pod is never looked up
+      # here. Node names (I21, I22, I25, I34-I36) are the fixture's
+      # instance IDs; cluset compacts them to I[21-22,25] and I[34-36].
+      # Switch order (S2 before S3) is deterministic (sorted breadth-first
+      # traversal of the fixture tree).
       healthCheckAsserts: |
         apiVersion: chainsaw.kyverno.io/v1alpha1
         kind: Test
@@ -211,7 +211,9 @@ spec:
                         namespace: slurm
                       data:
                         topology.conf: |
-                          SwitchName=SW1 Nodes=slinky-0
+                          SwitchName=S1 Switches=S[2-3]
+                          SwitchName=S2 Nodes=I[21-22,25]
+                          SwitchName=S3 Nodes=I[34-36]
 
     - name: slinky-slurm
       type: Helm

recipes/overlays/h100-kind-training-slurm.yaml

@@ -888,24 +888,24 @@ components:
     helm:
       defaultRepository: https://nvidia.github.io/topograph
       defaultChart: topograph/topograph
-      defaultVersion: "0.5.0"
+      defaultVersion: "1.0.0"
       defaultNamespace: topograph
     nodeScheduling:
       system:
         nodeSelectorPaths:
           - nodeSelector
-          - node-observer.nodeSelector
+          - nodeObserver.nodeSelector
         tolerationPaths:
           - tolerations
-          - node-observer.tolerations
+          - nodeObserver.tolerations
       accelerated:
         nodeSelectorPaths:
-          # node-observer.topograph.trigger.nodeSelector is a watch filter —
+          # nodeObserver.topograph.trigger.nodeSelector is a watch filter —
           # it narrows which node-change events trigger topology regeneration.
-          - node-observer.topograph.trigger.nodeSelector
-          - node-data-broker.nodeSelector
+          - nodeObserver.topograph.trigger.nodeSelector
+          - nodeDataBroker.nodeSelector
         tolerationPaths:
-          - node-data-broker.tolerations
+          - nodeDataBroker.tolerations
 
   - name: gatekeeper
     displayName: Gatekeeper