69 lines
1.5 KiB
YAML
69 lines
1.5 KiB
YAML
# Copyright (C) 2024 Intel Corporation
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
|
|
---
|
|
apiVersion: apps/v1
|
|
kind: Deployment
|
|
metadata:
|
|
name: tei-embedding-service-deploy
|
|
spec:
|
|
replicas: 1
|
|
selector:
|
|
matchLabels:
|
|
app: tei-embedding-service-deploy
|
|
template:
|
|
metadata:
|
|
annotations:
|
|
sidecar.istio.io/rewriteAppHTTPProbers: "true"
|
|
labels:
|
|
app: tei-embedding-service-deploy
|
|
spec:
|
|
containers:
|
|
- envFrom:
|
|
- configMapRef:
|
|
name: qna-config
|
|
image: ghcr.io/huggingface/text-embeddings-inference:cpu-1.2
|
|
name: tei-embedding-service-deploy
|
|
args:
|
|
- --model-id
|
|
- $(EMBEDDING_MODEL_ID)
|
|
volumeMounts:
|
|
- mountPath: /data
|
|
name: model-volume
|
|
- mountPath: /dev/shm
|
|
name: shm
|
|
ports:
|
|
- containerPort: 80
|
|
# resources:
|
|
# limits:
|
|
# cpu: 56000m
|
|
# memory: 26Gi
|
|
# requests:
|
|
# cpu: 56000m
|
|
# memory: 26Gi
|
|
serviceAccountName: default
|
|
volumes:
|
|
- name: model-volume
|
|
hostPath:
|
|
# directory location on host
|
|
path: /mnt/models
|
|
# this field is optional
|
|
type: Directory
|
|
- name: shm
|
|
emptyDir:
|
|
medium: Memory
|
|
sizeLimit: 1Gi
|
|
---
|
|
kind: Service
|
|
apiVersion: v1
|
|
metadata:
|
|
name: tei-embedding-svc
|
|
spec:
|
|
type: ClusterIP
|
|
selector:
|
|
app: tei-embedding-service-deploy
|
|
ports:
|
|
- name: service
|
|
port: 6006
|
|
targetPort: 80
|