Skip to content

Latest commit

 

History

History
162 lines (121 loc) · 4.99 KB

README.md

File metadata and controls

162 lines (121 loc) · 4.99 KB

SGLang Kubernetes Deployment Configurations

This directory contains Kubernetes Custom Resource Definition (CRD) templates for deploying SGLang inference graphs using the DynamoGraphDeployment resource.

Available Deployment Patterns

1. Aggregated Deployment (agg.yaml)

Basic deployment pattern with frontend and a single decode worker.

Architecture:

  • Frontend: OpenAI-compatible API server
  • SGLangDecodeWorker: Single worker handling both prefill and decode

2. Aggregated Router Deployment (agg_router.yaml)

Enhanced aggregated deployment with KV cache routing capabilities.

Architecture:

  • Frontend: OpenAI-compatible API server with router mode enabled (--router-mode kv)
  • SGLangDecodeWorker: Single worker handling both prefill and decode

3. Disaggregated Deployment (disagg.yaml)**

High-performance deployment with separated prefill and decode workers.

Architecture:

  • Frontend: HTTP API server coordinating between workers
  • SGLangDecodeWorker: Specialized decode-only worker (--disaggregation-mode decode)
  • SGLangPrefillWorker: Specialized prefill-only worker (--disaggregation-mode prefill)
  • Communication via NIXL transfer backend (--disaggregation-transfer-backend nixl)

CRD Structure

All templates use the DynamoGraphDeployment CRD:

apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
  name: <deployment-name>
spec:
  services:
    <ServiceName>:
      # Service configuration

Key Configuration Options

Resource Management:

resources:
  requests:
    cpu: "10"
    memory: "20Gi"
    gpu: "1"
  limits:
    cpu: "10"
    memory: "20Gi"
    gpu: "1"

Container Configuration:

extraPodSpec:
  mainContainer:
    image: my-registry/sglang-runtime:my-tag
    workingDir: /workspace/examples/backends/sglang
    args:
      - "python3"
      - "-m"
      - "dynamo.sglang"
      # Model-specific arguments

Prerequisites

Before using these templates, ensure you have:

  1. Dynamo Kubernetes Platform installed - See Installing Dynamo Kubernetes Platform
  2. Kubernetes cluster with GPU support
  3. Container registry access for SGLang runtime images
  4. HuggingFace token secret (referenced as envFromSecret: hf-token-secret)

Usage

1. Choose Your Template

Select the deployment pattern that matches your requirements:

  • Use agg.yaml for development/testing
  • Use agg_router.yaml for production with load balancing
  • Use disagg.yaml for maximum performance

2. Customize Configuration

Edit the template to match your environment:

# Update image registry and tag
image: my-registry/sglang-runtime:my-tag

# Configure your model
args:
  - "--model-path"
  - "your-org/your-model"
  - "--served-model-name"
  - "your-org/your-model"

3. Deploy

Use the following command to deploy the deployment file.

First, create a secret for the HuggingFace token.

export HF_TOKEN=your_hf_token
kubectl create secret generic hf-token-secret \
  --from-literal=HF_TOKEN=${HF_TOKEN} \
  -n ${NAMESPACE}

Then, deploy the model using the deployment file.

export DEPLOYMENT_FILE=agg.yaml
kubectl apply -f $DEPLOYMENT_FILE -n ${NAMESPACE}

4. Using Custom Dynamo Frameworks Image for SGLang

To use a custom dynamo frameworks image for SGLang, you can update the deployment file using yq:

export DEPLOYMENT_FILE=agg.yaml
export FRAMEWORK_RUNTIME_IMAGE=<sglang-image>

yq '.spec.services.[].extraPodSpec.mainContainer.image = env(FRAMEWORK_RUNTIME_IMAGE)' $DEPLOYMENT_FILE  > $DEPLOYMENT_FILE.generated
kubectl apply -f $DEPLOYMENT_FILE.generated -n $NAMESPACE

Model Configuration

All templates use DeepSeek-R1-Distill-Llama-8B as the default model. But you can use any sglang argument and configuration. Key parameters:

Monitoring and Health

  • Frontend health endpoint: http://<frontend-service>:8000/health
  • Liveness probes: Check process health every 60s

Further Reading

Troubleshooting

Common issues and solutions:

  1. Pod fails to start: Check image registry access and HuggingFace token secret
  2. GPU not allocated: Verify cluster has GPU nodes and proper resource limits
  3. Health check failures: Review model loading logs and increase initialDelaySeconds
  4. Out of memory: Increase memory limits or reduce model batch size

For additional support, refer to the deployment guide.