mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-08-01 16:24:44 +00:00
working now
This commit is contained in:
parent
1cb9d3bca2
commit
b63982ef00
6 changed files with 59 additions and 23 deletions
|
@ -21,7 +21,8 @@ export CODE_MODEL=bigcode/starcoder2-7b
|
|||
|
||||
# Set USE_EBS to false if you don't have permission to use EKS EBS
|
||||
export USE_EBS=${USE_EBS:-false}
|
||||
|
||||
set -euo pipefail
|
||||
set -x
|
||||
# HF_TOKEN should be set by the user; base64 encode it for the secret
|
||||
if [ -n "${HF_TOKEN:-}" ]; then
|
||||
export HF_TOKEN_BASE64=$(echo -n "$HF_TOKEN" | base64)
|
||||
|
@ -50,13 +51,35 @@ fi
|
|||
|
||||
|
||||
|
||||
# Apply the HF token secret if HF_TOKEN is provided
|
||||
if [ -n "${HF_TOKEN:-}" ]; then
|
||||
envsubst < ./set-secret.yaml.template | kubectl apply -f -
|
||||
fi
|
||||
# Create secrets and capture any errors
|
||||
echo "Creating Kubernetes secrets..."
|
||||
SECRET_OUTPUT=$(envsubst < ./set-secret.yaml.template | kubectl apply -f - 2>&1) || {
|
||||
echo "ERROR: Failed to create secrets. Error output:"
|
||||
echo "$SECRET_OUTPUT"
|
||||
exit 1
|
||||
}
|
||||
|
||||
set -euo pipefail
|
||||
set -x
|
||||
# Wait a moment for Kubernetes to process the secrets
|
||||
echo "Waiting for secrets to be processed..."
|
||||
sleep 2
|
||||
|
||||
# Verify that the secrets were created successfully
|
||||
echo "Verifying that secrets were created successfully..."
|
||||
|
||||
# Check each secret with better error handling
|
||||
for SECRET_NAME in "hf-token-secret" "ngc-docker-registry" "ngc-api"; do
|
||||
echo "Checking for secret: $SECRET_NAME"
|
||||
if ! kubectl get secret "$SECRET_NAME" &> /dev/null; then
|
||||
echo "ERROR: Secret '$SECRET_NAME' not found in the cluster."
|
||||
echo "Secret creation output was:"
|
||||
echo "$SECRET_OUTPUT"
|
||||
echo "Current secrets in the namespace:"
|
||||
kubectl get secrets
|
||||
exit 1
|
||||
fi
|
||||
done
|
||||
|
||||
echo "Secret verification successful. All required secrets are present."
|
||||
# Apply templates with appropriate storage configuration based on USE_EBS setting
|
||||
if [ "$USE_EBS" = "true" ]; then
|
||||
echo "Using EBS storage for persistent volumes"
|
||||
|
@ -73,6 +96,7 @@ if [ "$USE_EBS" = "true" ]; then
|
|||
envsubst < ./stack-k8s.yaml.template | kubectl apply -f -
|
||||
envsubst < ./ingress-k8s.yaml.template | kubectl apply -f -
|
||||
envsubst < ./ui-k8s.yaml.template | kubectl apply -f -
|
||||
envsubst < ./ui-service-k8s.yaml.template | kubectl apply -f -
|
||||
else
|
||||
echo "Using emptyDir for storage (data will not persist across pod restarts)"
|
||||
# Process templates to replace EBS storage with emptyDir
|
||||
|
@ -90,4 +114,5 @@ else
|
|||
envsubst < ./stack-k8s.yaml.template | sed 's/persistentVolumeClaim:/emptyDir: {}/g' | sed '/claimName:/d' | kubectl apply -f -
|
||||
envsubst < ./ingress-k8s.yaml.template | sed 's/persistentVolumeClaim:/emptyDir: {}/g' | sed '/claimName:/d' | kubectl apply -f -
|
||||
envsubst < ./ui-k8s.yaml.template | sed 's/persistentVolumeClaim:/emptyDir: {}/g' | sed '/claimName:/d' | kubectl apply -f -
|
||||
envsubst < ./ui-service-k8s.yaml.template | kubectl apply -f -
|
||||
fi
|
||||
|
|
|
@ -6,12 +6,9 @@ spec:
|
|||
type: LoadBalancer
|
||||
selector:
|
||||
app.kubernetes.io/name: llama-stack
|
||||
app.kubernetes.io/component: server
|
||||
ports:
|
||||
- name: llama-stack-api
|
||||
port: 8321
|
||||
targetPort: 8321
|
||||
protocol: TCP
|
||||
- name: llama-stack-ui
|
||||
port: 8322
|
||||
targetPort: 8322
|
||||
protocol: TCP
|
||||
|
|
|
@ -1,3 +1,6 @@
|
|||
# -------------------------------------------------
|
||||
# Hugging Face Token Secret
|
||||
# -------------------------------------------------
|
||||
apiVersion: v1
|
||||
kind: Secret
|
||||
metadata:
|
||||
|
@ -5,24 +8,22 @@ metadata:
|
|||
type: Opaque
|
||||
data:
|
||||
token: ${HF_TOKEN_BASE64}
|
||||
|
||||
---
|
||||
# -------------------------------------------------
|
||||
# NGC Docker Registry Secret
|
||||
# -------------------------------------------------
|
||||
|
||||
apiVersion: apps/v1
|
||||
apiVersion: v1
|
||||
kind: Secret
|
||||
metadata:
|
||||
name: ngc-docker-registry
|
||||
type: kubernetes.io/dockerconfigjson
|
||||
data:
|
||||
.dockerconfigjson: ${NGC_DOCKER_CONFIG_JSON}
|
||||
|
||||
---
|
||||
# -------------------------------------------------
|
||||
# NGC API Secret
|
||||
# -------------------------------------------------
|
||||
|
||||
apiVersion: apps/v1
|
||||
apiVersion: v1
|
||||
kind: Secret
|
||||
metadata:
|
||||
name: ngc-api
|
||||
|
|
|
@ -34,17 +34,16 @@ spec:
|
|||
command: ['sh', '-c', 'until nc -z llm-nim-code.default.svc.cluster.local 8000; do echo waiting for llm-nim-code on port 8001; sleep 2; done;']
|
||||
containers:
|
||||
- name: llama-stack
|
||||
image: llamastack/distribution-starter:latest
|
||||
imagePullPolicy: Always # since we have specified latest instead of a version
|
||||
image: llamastack/distribution-starter:0.2.15
|
||||
resources:
|
||||
requests:
|
||||
memory: "512Mi"
|
||||
cpu: "500m"
|
||||
ephemeral-storage: "2Gi"
|
||||
ephemeral-storage: "6Gi"
|
||||
limits:
|
||||
memory: "1Gi"
|
||||
cpu: "1000m"
|
||||
ephemeral-storage: "5Gi"
|
||||
ephemeral-storage: "6Gi"
|
||||
env:
|
||||
- name: ENABLE_CHROMADB
|
||||
value: "true"
|
||||
|
|
|
@ -35,7 +35,7 @@ spec:
|
|||
apt-get update && apt-get install -y git
|
||||
# Clone the repository
|
||||
git clone https://github.com/meta-llama/llama-stack.git /app
|
||||
|
||||
git checkout k8s_demo
|
||||
# Navigate to the playground directory
|
||||
cd /app/llama_stack/distribution/ui
|
||||
|
||||
|
@ -45,4 +45,4 @@ spec:
|
|||
# Run the Streamlit app
|
||||
streamlit run app.py --server.port=8322 --server.address=0.0.0.0
|
||||
ports:
|
||||
- containerPort: 8501
|
||||
- containerPort: 8322
|
||||
|
|
14
docs/source/distributions/k8s/ui-service-k8s.yaml.template
Normal file
14
docs/source/distributions/k8s/ui-service-k8s.yaml.template
Normal file
|
@ -0,0 +1,14 @@
|
|||
apiVersion: v1
|
||||
kind: Service
|
||||
metadata:
|
||||
name: llama-stack-ui-service
|
||||
spec:
|
||||
type: LoadBalancer
|
||||
selector:
|
||||
app.kubernetes.io/name: llama-stack
|
||||
app.kubernetes.io/component: ui
|
||||
ports:
|
||||
- name: llama-stack-ui
|
||||
port: 8322
|
||||
targetPort: 8322
|
||||
protocol: TCP
|
Loading…
Add table
Add a link
Reference in a new issue