feature: 对接deer-flow (#54)

feature: 对接deer-flow
2025-11-04 20:30:40 +08:00
parent dc30b0d892
commit f3958f08d9
710 changed files with 112812 additions and 52 deletions
--- a/deployment/helm/milvus/charts/tei/.helmignore
+++ b/deployment/helm/milvus/charts/tei/.helmignore
@@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/
--- a/deployment/helm/milvus/charts/tei/Chart.yaml
+++ b/deployment/helm/milvus/charts/tei/Chart.yaml
@@ -0,0 +1,16 @@
+apiVersion: v1
+appVersion: 1.6.0
+description: A text-embeddings-inference Helm chart for Kubernetes
+home: https://github.com/huggingface/text-embeddings-inference
+icon: https://avatars.githubusercontent.com/u/25720743?s=200&v=4
+keywords:
+- embeddings
+- inference
+maintainers:
+- email: devops@zilliz.com
+  name: zilliz
+name: tei
+sources:
+- https://github.com/huggingface/text-embeddings-inference
+type: application
+version: 1.6.0
--- a/deployment/helm/milvus/charts/tei/README.md
+++ b/deployment/helm/milvus/charts/tei/README.md
@@ -0,0 +1,169 @@
+# Text Embeddings Inference (TEI) Integration Guide
+
+This document describes how to use Text Embeddings Inference (TEI) service with Milvus Helm Chart, and how to integrate TEI with Milvus. TEI is an open-source project developed by Hugging Face, available at [https://github.com/huggingface/text-embeddings-inference](https://github.com/huggingface/text-embeddings-inference).
+
+## Overview
+
+Text Embeddings Inference (TEI) is a high-performance text embedding model inference service that converts text into vector representations. Milvus is a vector database that can store and retrieve these vectors. By combining the two, you can build powerful semantic search and retrieval systems.
+
+## Deployment Methods
+
+This guide provides two ways to use TEI:
+1. Deploy TEI service directly through the Milvus Helm Chart
+2. Use external TEI service with Milvus integration
+
+## Deploy TEI through Milvus Helm Chart
+
+### Basic Configuration
+
+```yaml
+modelId: "BAAI/bge-large-en-v1.5"  # Specify the model to use
+```
+
+This is the simplest configuration, just specify `enabled: true` and the desired `modelId`.
+
+### Complete Configuration Options
+
+```yaml
+modelId: "BAAI/bge-large-en-v1.5"    # Model ID
+extraArgs: []                        # Additional command line arguments for TEI, such as "--max-batch-tokens=16384", "--max-client-batch-size=32", "--max-concurrent-requests=128", etc.
+replicaCount: 1                      # Number of TEI replicas
+image:
+  repository: ghcr.io/huggingface/text-embeddings-inference  # Image repository
+  tag: cpu-1.6                       # Image tag (CPU version)
+  pullPolicy: IfNotPresent           # Image pull policy
+service:
+  type: ClusterIP                    # Service type
+  port: 8080                         # Service port
+  annotations: {}                    # Service annotations
+  labels: {}                         # Service labels
+resources:                           # Resource configuration
+  requests:
+    cpu: "4"                         # CPU request
+    memory: "8Gi"                    # Memory request
+  limits:
+    cpu: "8"                         # CPU limit
+    memory: "16Gi"                   # Memory limit
+persistence:                         # Persistence storage configuration
+  enabled: true                      # Enable persistence storage
+  mountPath: "/data"                 # Mount path
+  annotations: {}                    # Storage annotations
+  persistentVolumeClaim:             # PVC configuration
+    existingClaim: ""                # Use existing PVC
+    storageClass:                    # Storage class
+    accessModes: ReadWriteOnce       # Access modes
+    size: 50Gi                       # Storage size
+    subPath: ""                      # Sub path
+nodeSelector: {}                     # Node selector
+affinity: {}                         # Affinity configuration
+tolerations: []                      # Tolerations
+topologySpreadConstraints: []        # Topology spread constraints
+extraEnv: []                         # Additional environment variables
+```
+
+### Using GPU Acceleration
+
+If you have GPU resources, you can use the GPU version of the TEI image to accelerate inference:
+
+```yaml
+enabled: true
+modelId: "BAAI/bge-large-en-v1.5"
+image:
+  repository: ghcr.io/huggingface/text-embeddings-inference
+  tag: 1.6  # GPU version
+resources:
+  limits:
+    nvidia.com/gpu: 1  # Allocate 1 GPU
+```
+
+
+## Frequently Asked Questions
+
+### How to determine the embedding dimension of a model?
+
+Different models have different embedding dimensions. Here are the dimensions of some commonly used models:
+- BAAI/bge-large-en-v1.5: 1024
+- BAAI/bge-base-en-v1.5: 768
+- nomic-ai/nomic-embed-text-v1: 768
+- sentence-transformers/all-mpnet-base-v2: 768
+
+You can find this information in the model's documentation or get it through the TEI service's API.
+
+### How to test if the TEI service is working properly?
+
+After deploying the TEI service, you can use the following commands to test if the service is working properly:
+
+```bash
+# Get the TEI service endpoint
+export TEI_SERVICE=$(kubectl get svc -l component=text-embeddings-inference -o jsonpath='{.items[0].metadata.name}')
+
+# Test the embedding functionality
+kubectl run -it --rm curl --image=curlimages/curl -- curl -X POST "http://${TEI_SERVICE}:8080/embed" \
+  -H "Content-Type: application/json" \
+  -d '{"inputs":"This is a test text"}'
+```
+
+### How to use TEI-generated embeddings in Milvus?
+
+In Milvus, you can use TEI-generated embeddings for the following operations:
+
+1. When creating a collection, specify the vector dimension to match the TEI model output dimension
+2. Before inserting data, use the TEI service to convert text to vectors
+3. When searching, similarly use the TEI service to convert query text to vectors
+
+## Using Milvus Text Embedding Function
+
+Milvus provides a text embedding function feature that allows you to generate vector embeddings directly within Milvus. You can configure Milvus to use TEI as the backend for this function.
+
+### Using the Text Embedding Function in Milvus
+
+1. Specify the embedding function when creating a collection:
+
+```python
+from pymilvus import connections, Collection, FieldSchema, CollectionSchema, DataType
+
+# Connect to Milvus
+connections.connect(host="localhost", port="19530")
+
+# Define collection schema
+fields = [
+    FieldSchema(name="id", dtype=DataType.INT64, is_primary=True),
+    FieldSchema(name="text", dtype=DataType.VARCHAR, max_length=1000),
+    FieldSchema(name="vector", dtype=DataType.FLOAT_VECTOR, dim=768)  # Dimension should match model output
+]
+schema = CollectionSchema(fields=fields, description="Text collection with embedding function")
+
+# Create collection and specify embedding function
+collection = Collection(
+    name="text_collection",
+    schema=schema,
+    embedding_field="text",  # Specify the field to embed
+    vector_field="vector",   # Specify the field to store embedding vectors
+    embedding_config={
+        "provider": "tei",
+        "model_id": "BAAI/bge-large-en-v1.5",
+        "endpoint": "http://tei-service:8080"
+    }
+)
+```
+
+2. Automatically generate embeddings when inserting data:
+
+```python
+# Insert data, Milvus will automatically call the TEI service to generate embedding vectors
+collection.insert([
+    {"id": 1, "text": "This is a sample document about artificial intelligence."},
+    {"id": 2, "text": "Vector databases are designed to handle embeddings efficiently."}
+])
+```
+
+3. Automatically generate query embeddings when searching:
+
+```python
+# Search directly using text, Milvus will automatically call the TEI service to generate query vectors
+results = collection.search(
+    query_texts=["Tell me about AI technology"],
+    embedding_field="text",
+    limit=3
+)
+```
--- a/deployment/helm/milvus/charts/tei/ci/tei-mininum-values.yaml
+++ b/deployment/helm/milvus/charts/tei/ci/tei-mininum-values.yaml
@@ -0,0 +1,12 @@
+resources:
+  requests:
+    cpu: "0.1"
+    memory: "2Gi"
+  limits:
+    cpu: "8"
+    memory: "16Gi"
+
+persistence:
+  enabled: true
+  persistentVolumeClaim:
+    size: 10Gi
--- a/deployment/helm/milvus/charts/tei/templates/_helpers.tpl
+++ b/deployment/helm/milvus/charts/tei/templates/_helpers.tpl
@@ -0,0 +1,62 @@
+{{/*
+Expand the name of the chart.
+*/}}
+{{- define "tei.name" -}}
+{{- default .Chart.Name .Values.nameOverride | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Create a default fully qualified app name.
+We truncate at 63 chars because some Kubernetes name fields are limited to this (by the DNS naming spec).
+If release name contains chart name it will be used as a full name.
+*/}}
+{{- define "tei.fullname" -}}
+{{- if .Values.fullnameOverride }}
+{{- .Values.fullnameOverride | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- $name := default .Chart.Name .Values.nameOverride }}
+{{- if contains $name .Release.Name }}
+{{- .Release.Name | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- printf "%s-%s" .Release.Name $name | trunc 63 | trimSuffix "-" }}
+{{- end }}
+{{- end }}
+{{- end }}
+
+{{/*
+Create chart name and version as used by the chart label.
+*/}}
+{{- define "tei.chart" -}}
+{{- printf "%s-%s" .Chart.Name .Chart.Version | replace "+" "_" | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Common labels
+*/}}
+{{- define "tei.labels" -}}
+helm.sh/chart: {{ include "tei.chart" . }}
+{{ include "tei.selectorLabels" . }}
+{{- if .Chart.AppVersion }}
+app.kubernetes.io/version: {{ .Chart.AppVersion | quote }}
+{{- end }}
+app.kubernetes.io/managed-by: {{ .Release.Service }}
+{{- end }}
+
+{{/*
+Selector labels
+*/}}
+{{- define "tei.selectorLabels" -}}
+app.kubernetes.io/name: {{ include "tei.name" . }}
+app.kubernetes.io/instance: {{ .Release.Name }}
+{{- end }}
+
+{{/*
+Create the name of the service account to use
+*/}}
+{{- define "tei.serviceAccountName" -}}
+{{- if .Values.serviceAccount.create }}
+{{- default (include "tei.fullname" .) .Values.serviceAccount.name }}
+{{- else }}
+{{- default "default" .Values.serviceAccount.name }}
+{{- end }}
+{{- end }}
--- a/deployment/helm/milvus/charts/tei/templates/headless-service.yaml
+++ b/deployment/helm/milvus/charts/tei/templates/headless-service.yaml
@@ -0,0 +1,15 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ include "tei.fullname" . }}-headless
+  labels:
+    {{- include "tei.labels" . | nindent 4 }}
+spec:
+  clusterIP: None
+  ports:
+  - port: {{ .Values.service.port }}
+    targetPort: http
+    protocol: TCP
+    name: http
+  selector:
+    {{- include "tei.selectorLabels" . | nindent 4 }}
--- a/deployment/helm/milvus/charts/tei/templates/service.yaml
+++ b/deployment/helm/milvus/charts/tei/templates/service.yaml
@@ -0,0 +1,19 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ include "tei.fullname" . }}
+  labels:
+    {{- include "tei.labels" . | nindent 4 }}
+  {{- with .Values.service.annotations }}
+  annotations:
+    {{- toYaml . | nindent 4 }}
+  {{- end }}
+spec:
+  type: {{ .Values.service.type }}
+  ports:
+  - port: {{ .Values.service.port }}
+    targetPort: http
+    protocol: TCP
+    name: http
+  selector:
+    {{- include "tei.selectorLabels" . | nindent 4 }}
--- a/deployment/helm/milvus/charts/tei/templates/serviceaccount.yaml
+++ b/deployment/helm/milvus/charts/tei/templates/serviceaccount.yaml
@@ -0,0 +1,13 @@
+{{- if .Values.serviceAccount.create -}}
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: {{ include "tei.serviceAccountName" . }}
+  labels:
+    {{- include "tei.labels" . | nindent 4 }}
+  {{- with .Values.serviceAccount.annotations }}
+  annotations:
+    {{- toYaml . | nindent 4 }}
+  {{- end }}
+automountServiceAccountToken: {{ .Values.serviceAccount.automount }}
+{{- end }}
--- a/deployment/helm/milvus/charts/tei/templates/statefulset.yaml
+++ b/deployment/helm/milvus/charts/tei/templates/statefulset.yaml
@@ -0,0 +1,103 @@
+apiVersion: apps/v1
+kind: StatefulSet
+metadata:
+  name: {{ include "tei.fullname" . }}
+  labels:
+    {{- include "tei.labels" . | nindent 4 }}
+    {{- if .Values.labels }}
+      {{ toYaml .Values.labels | indent 4 }}
+    {{- end }}
+  {{- if .Values.annotations }}
+  annotations:
+    {{ toYaml .Values.annotations | indent 4 }}
+  {{- end }}
+spec:
+  serviceName: {{ template "tei.fullname" . }}-headless
+  podManagementPolicy: Parallel
+  replicas: {{ .Values.replicaCount | default 1 }}
+  selector:
+    matchLabels:
+      {{- include "tei.selectorLabels" . | nindent 6 }}
+  template:
+    metadata:
+      {{- with .Values.podAnnotations }}
+      annotations:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      labels:
+        {{- include "tei.labels" . | nindent 8 }}
+        {{- with .Values.podLabels }}
+        {{- toYaml . | nindent 8 }}
+        {{- end }}
+    spec:
+      {{- with .Values.imagePullSecrets }}
+      imagePullSecrets:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      serviceAccountName: {{ include "tei.serviceAccountName" . }}
+      securityContext:
+        {{- toYaml .Values.podSecurityContext | nindent 8 }}
+      containers:
+      - name: {{ .Chart.Name }}
+        securityContext:
+          {{- toYaml .Values.securityContext | nindent 12 }}
+        image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
+        imagePullPolicy: {{ .Values.image.pullPolicy }}
+        args:
+        - "--model-id"
+        - "{{ .Values.modelId }}"
+        {{- range .Values.extraArgs }}
+        - {{ . | quote }}
+        {{- end }}
+        ports:
+        - containerPort: 80
+          name: http
+        resources:
+          {{- toYaml .Values.resources | nindent 12 }}
+        {{- if .Values.extraEnv }}
+        env:
+          {{ toYaml .Values.extraEnv | indent 8 }}
+        {{- end }}
+        volumeMounts:
+        - name: data-volume
+          mountPath: {{ .Values.persistence.mountPath }}
+        {{- if .Values.volumeMounts }}
+          {{ toYaml .Values.volumeMounts | indent 8 }}
+        {{- end }}
+      volumes:
+      {{- if not .Values.persistence.enabled }}
+      - name: data-volume
+        emptyDir: {}
+      {{- end }}
+      {{- with .Values.volumes }}
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.nodeSelector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.affinity }}
+      affinity:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.tolerations }}
+      tolerations:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+  {{- if .Values.persistence.enabled }}
+  volumeClaimTemplates:
+  - metadata:
+      name: data-volume
+    spec:
+      accessModes: [ "ReadWriteOnce" ]
+      {{- if .Values.persistence.persistentVolumeClaim.storageClass }}
+      {{- if (eq "-" .Values.persistence.persistentVolumeClaim.storageClass) }}
+      storageClassName: ""
+      {{- else }}
+      storageClassName: "{{ .Values.persistence.persistentVolumeClaim.storageClass }}"
+      {{- end }}
+      {{- end }}
+      resources:
+        requests:
+          storage: {{ .Values.persistence.persistentVolumeClaim.size }}
+  {{- end }}
--- a/deployment/helm/milvus/charts/tei/values.yaml
+++ b/deployment/helm/milvus/charts/tei/values.yaml
@@ -0,0 +1,85 @@
+# Default values for tei.
+# This is a YAML-formatted file.
+# Declare variables to be passed into your templates.
+
+# TEI model configuration
+modelId: "BAAI/bge-large-en-v1.5"
+# Additional TEI configuration
+extraArgs: []
+
+replicaCount: 1
+
+image:
+  repository: ghcr.io/huggingface/text-embeddings-inference
+  tag: cpu-1.6
+  pullPolicy: IfNotPresent
+imagePullSecrets: []
+nameOverride: ""
+fullnameOverride: ""
+
+serviceAccount:
+  # Specifies whether a service account should be created
+  create: true
+  # Automatically mount a ServiceAccount's API credentials?
+  automount: true
+  # Annotations to add to the service account
+  annotations: {}
+  # The name of the service account to use.
+  # If not set and create is true, a name is generated using the fullname template
+  name: ""
+
+podAnnotations: {}
+podLabels: {}
+
+podSecurityContext: {}
+  # fsGroup: 2000
+
+securityContext: {}
+  # capabilities:
+  #   drop:
+  #   - ALL
+  # readOnlyRootFilesystem: true
+  # runAsNonRoot: true
+  # runAsUser: 1000
+
+service:
+  type: ClusterIP
+  port: 8080
+  annotations: {}
+  labels: {}
+resources:
+  requests:
+    cpu: "4"
+    memory: "8Gi"
+  limits:
+    cpu: "8"
+    memory: "16Gi"
+persistence:
+  enabled: true
+  mountPath: "/data"
+  annotations: {}
+  persistentVolumeClaim:
+    existingClaim: ""
+    storageClass:
+    accessModes: ReadWriteOnce
+    size: 50Gi
+    subPath: ""
+nodeSelector: {}
+affinity: {}
+tolerations: []
+topologySpreadConstraints: []
+extraEnv: []
+
+
+# Additional volumes on the output Deployment definition.
+volumes: []
+# - name: foo
+#   secret:
+#     secretName: mysecret
+#     optional: false
+
+# Additional volumeMounts on the output Deployment definition.
+volumeMounts: []
+# - name: foo
+#   mountPath: "/etc/foo"
+#   readOnly: true