Skip to content

[Bugfix]: StatefulSet readiness detection and add Dynamo demo video #856

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
rootfs merged 2 commits into from
Dec 17, 2025
+38 −4
Merged

[Bugfix]: StatefulSet readiness detection and add Dynamo demo video #856

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b3bba04
fix: StatefulSet readiness detection and add Dynamo demo video
abdallahsamabd Dec 17, 2025
c6cae4f
Merge branch 'main' into dynamo/demo
abdallahsamabd Dec 17, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 35 additions & 2 deletions deploy/kubernetes/dynamo/dynamo-resources/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,10 @@

This directory contains Kubernetes manifests for E2E testing of **Semantic Router with NVIDIA Dynamo integration**.

## 🎥 Demo Video

▶️ **[Watch the E2E Demo on YouTube](https://www.youtube.com/watch?v=rRULSR9gTds&list=PLmrddZ45wYcuPrXisC-yl7bMI39PLo4LO&index=2)**

## ⚠️ GPU Requirements

**This test requires a VM with at least 3 GPUs:**
Expand All @@ -12,13 +16,42 @@ This directory contains Kubernetes manifests for E2E testing of **Semantic Route
| VLLMPrefillWorker | GPU 1 | Handles prefill phase of inference |
| VLLMDecodeWorker | GPU 2 | Handles decode phase of inference |

## 🔧 Prerequisites (One-Time Setup)

> **Tested on:** RHEL 9 VM with NVIDIA GPUs

Before running the E2E tests, you must configure Docker to use the NVIDIA runtime as the default. This is a **one-time setup** that persists across reboots.

### Step 1: Configure NVIDIA Runtime as Default

```bash
sudo nvidia-ctk runtime configure --runtime=docker --set-as-default
```

### Step 2: Restart Docker

```bash
sudo systemctl restart docker
```

### Step 3: Verify Configuration

```bash
docker info | grep -i "default runtime"
```

You should see: `Default Runtime: nvidia`

> **Note:** The E2E framework verifies this configuration but does not set it automatically (requires sudo privileges).

## What the E2E Framework Does

The E2E framework automatically:

1. Sets Docker runtime to `nvidia` (required for GPU passthrough)
1. **Verifies** Docker runtime is set to `nvidia` (fails with instructions if not configured)
2. Creates Kind cluster with GPU support
3. Copies NVIDIA libraries to the Kind worker node
4. Deploys the NVIDIA device plugin
5. Restores Docker runtime to default after tests complete

## What We Test

Expand Down
5 changes: 3 additions & 2 deletions e2e/profiles/dynamo/profile.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -594,8 +594,9 @@ func (p *Profile) waitForStatefulSet(ctx context.Context, kubeConfig, namespace,
defer cancel()

timeoutSeconds := int(timeout.Seconds())
cmd := exec.CommandContext(ctx, "kubectl", "wait",
"--for=condition=Ready",
// Use rollout status for StatefulSets instead of wait --for=condition=Ready
// (StatefulSets don't have a Ready condition like Deployments)
cmd := exec.CommandContext(ctx, "kubectl", "rollout", "status",
fmt.Sprintf("statefulset/%s", name),
"-n", namespace,
fmt.Sprintf("--timeout=%ds", timeoutSeconds),
Expand Down
Loading