feat: Support for Qdrant vector search

backend/apps/northbound_knowledge_app.py

@@ -6,7 +6,7 @@
 from fastapi import APIRouter, Body, File, Form, Path, Path as PathParam, Query, Request, HTTPException, UploadFile
 from fastapi.responses import JSONResponse, RedirectResponse, StreamingResponse
 
-from consts.const import ASSET_OWNER_TENANT_ID, VectorDatabaseType
+from consts.const import ASSET_OWNER_TENANT_ID
 from consts.exceptions import (
     LimitExceededError,
     UnauthorizedError,
@@ -65,7 +65,7 @@ async def get_list_indices(
     """
     try:
         ctx = await _require_asset_owner_context(request)
-        vdb_core = get_vector_db_core(db_type=VectorDatabaseType.ELASTICSEARCH)
+        vdb_core = get_vector_db_core()
         return ElasticSearchService.list_indices(
             pattern, True, ctx.tenant_id, ctx.user_id, vdb_core
         )
@@ -111,7 +111,7 @@ async def create_new_index(
     """
     try:
         ctx = await _require_asset_owner_context(request)
-        vdb_core = get_vector_db_core(db_type=VectorDatabaseType.ELASTICSEARCH)
+        vdb_core = get_vector_db_core()
 
         ingroup_permission = None
         group_ids = None
@@ -163,7 +163,7 @@ async def delete_index(
     logger.debug("Received northbound request to delete knowledge base")
     try:
         ctx = await _require_asset_owner_context(request)
-        vdb_core = get_vector_db_core(db_type=VectorDatabaseType.ELASTICSEARCH)
+        vdb_core = get_vector_db_core()
         return await ElasticSearchService.full_delete_knowledge_base(
             index_name, vdb_core, ctx.user_id
         )
@@ -195,7 +195,7 @@ async def get_index_files(
     """
     try:
         ctx = await _require_asset_owner_context(request)
-        vdb_core = get_vector_db_core(db_type=VectorDatabaseType.ELASTICSEARCH)
+        vdb_core = get_vector_db_core()
         logger.debug(
             "Listing files for index %s, tenant_id=%s, user_id=%s",
             index_name,
@@ -370,7 +370,7 @@ async def delete_documents(
     """Delete a document by scope. Restricted to asset administrators."""
     try:
         await _require_asset_owner_context(request)
-        vdb_core = get_vector_db_core(db_type=VectorDatabaseType.ELASTICSEARCH)
+        vdb_core = get_vector_db_core()
         logger.debug(
             "Deleting documents for index %s scope=%s", index_name, scope
         )

backend/consts/const.py

@@ -19,15 +19,22 @@
 class VectorDatabaseType(str, Enum):
     ELASTICSEARCH = "elasticsearch"
     DATAMATE = "datamate"
+    QDRANT = "qdrant"
 
 
 # Elasticsearch Configuration
+VECTOR_DATABASE_TYPE = os.getenv("VECTOR_DATABASE_TYPE", VectorDatabaseType.ELASTICSEARCH.value)
 ES_HOST = os.getenv("ELASTICSEARCH_HOST")
 ES_API_KEY = os.getenv("ELASTICSEARCH_API_KEY")
 ES_PASSWORD = os.getenv("ELASTIC_PASSWORD")
 ES_USERNAME = "elastic"
 ELASTICSEARCH_SERVICE = os.getenv("ELASTICSEARCH_SERVICE")
 
+# Qdrant
+QDRANT_URL = os.getenv("QDRANT_URL")
+QDRANT_API_KEY = os.getenv("QDRANT_API_KEY")
+QDRANT_TIMEOUT = float(os.getenv("QDRANT_TIMEOUT", "20"))
+
 # Data Processing Service Configuration
 DATA_PROCESS_SERVICE = os.getenv("DATA_PROCESS_SERVICE")
 CLIP_MODEL_PATH = os.getenv("CLIP_MODEL_PATH")

backend/services/vectordatabase_service.py

@@ -26,6 +26,7 @@
 from nexent.vector_database.base import VectorDatabaseCore
 from nexent.vector_database.elasticsearch_core import ElasticSearchCore
 from nexent.vector_database.datamate_core import DataMateCore
+from nexent.vector_database.qdrant_core import QdrantCore
 
 from consts.const import (
     ASSET_OWNER_TENANT_ID,
@@ -37,6 +38,10 @@
     LANGUAGE,
     PERMISSION_EDIT,
     PERMISSION_READ,
+    QDRANT_API_KEY,
+    QDRANT_TIMEOUT,
+    QDRANT_URL,
+    VECTOR_DATABASE_TYPE,
     VectorDatabaseType,
 )
 from consts.model import ChunkCreateRequest, ChunkUpdateRequest
@@ -237,7 +242,7 @@ def get_embedding_model_by_index_name(tenant_id: str, index_name: str) -> tuple[
 
 
 def get_vector_db_core(
-    db_type: VectorDatabaseType = VectorDatabaseType.ELASTICSEARCH, tenant_id: Optional[str] = None,
+    db_type: Optional[VectorDatabaseType] = None, tenant_id: Optional[str] = None,
 ) -> VectorDatabaseCore:
     """
     Return a VectorDatabaseCore implementation based on the requested type.
@@ -252,15 +257,24 @@ def get_vector_db_core(
     Raises:
         ValueError: If the requested database type is not supported.
     """
-    if db_type == VectorDatabaseType.ELASTICSEARCH:
+    selected_db_type = db_type or VectorDatabaseType(VECTOR_DATABASE_TYPE)
+
+    if selected_db_type == VectorDatabaseType.ELASTICSEARCH:
         return ElasticSearchCore(
             host=ES_HOST,
             api_key=ES_API_KEY,
             verify_certs=False,
             ssl_show_warn=False,
         )
 
-    if db_type == VectorDatabaseType.DATAMATE:
+    if selected_db_type == VectorDatabaseType.QDRANT:
+        return QdrantCore(
+            url=QDRANT_URL,
+            api_key=QDRANT_API_KEY,
+            timeout=QDRANT_TIMEOUT,
+        )
+
+    if selected_db_type == VectorDatabaseType.DATAMATE:
         if tenant_id:
             datamate_url = tenant_config_manager.get_app_config(
                 DATAMATE_URL, tenant_id=tenant_id)
@@ -271,7 +285,7 @@ def get_vector_db_core(
         else:
             raise ValueError("tenant_id must be provided for DataMate")
 
-    raise ValueError(f"Unsupported vector database type: {db_type}")
+    raise ValueError(f"Unsupported vector database type: {selected_db_type}")
 
 
 def _rethrow_or_plain(exc: Exception) -> None:

deploy/common/common.sh

@@ -1595,7 +1595,7 @@ deployment_compute_docker_ports() {
   fi
 
   if deployment_csv_contains "$DEPLOYMENT_COMPONENTS" "infrastructure"; then
-    ports+=(9210 9310 5434 6379 9010 9011)
+    ports+=(9210 9310 6333 5434 6379 9010 9011)
   fi
   if deployment_csv_contains "$DEPLOYMENT_COMPONENTS" "application"; then
     ports+=(5010 5014 5011 5015 5013 3000)
@@ -1674,6 +1674,7 @@ deployment_apply_image_source() {
   export NEXENT_DATA_PROCESS_IMAGE="${NEXENT_DATA_PROCESS_IMAGE:-nexent/nexent-data-process:$version}"
   export NEXENT_MCP_DOCKER_IMAGE="${NEXENT_MCP_DOCKER_IMAGE:-nexent/nexent-mcp:$version}"
   export ELASTICSEARCH_IMAGE="${ELASTICSEARCH_IMAGE:-docker.elastic.co/elasticsearch/elasticsearch:8.17.4}"
+  export QDRANT_IMAGE="${QDRANT_IMAGE:-qdrant/qdrant:v1.18.2}"
   export POSTGRESQL_IMAGE="${POSTGRESQL_IMAGE:-postgres:15-alpine}"
   export REDIS_IMAGE="${REDIS_IMAGE:-redis:alpine}"
   export MINIO_IMAGE="${MINIO_IMAGE:-quay.io/minio/minio:RELEASE.2023-12-20T01-00-02Z}"
@@ -1787,6 +1788,7 @@ deployment_render_docker_env() {
     printf 'NEXENT_DATA_PROCESS_IMAGE="%s"\n' "$NEXENT_DATA_PROCESS_IMAGE"
     printf 'NEXENT_MCP_DOCKER_IMAGE="%s"\n' "$NEXENT_MCP_DOCKER_IMAGE"
     printf 'ELASTICSEARCH_IMAGE="%s"\n' "$ELASTICSEARCH_IMAGE"
+    printf 'QDRANT_IMAGE="%s"\n' "$QDRANT_IMAGE"
     printf 'POSTGRESQL_IMAGE="%s"\n' "$POSTGRESQL_IMAGE"
     printf 'REDIS_IMAGE="%s"\n' "$REDIS_IMAGE"
     printf 'MINIO_IMAGE="%s"\n' "$MINIO_IMAGE"

deploy/docker/compose/docker-compose.prod.yml

@@ -48,6 +48,20 @@ services:
         max-size: "100m"  # Maximum size of a single log file
         max-file: "3"     # Maximum number of log files to keep
 
+  nexent-qdrant:
+    image: ${QDRANT_IMAGE}
+    container_name: nexent-qdrant
+    volumes:
+      - ${ROOT_DIR}/qdrant:/qdrant/storage
+    networks:
+      - nexent
+    restart: always
+    logging:
+      driver: "json-file"
+      options:
+        max-size: "100m"
+        max-file: "3"
+
   nexent-postgresql:
     image: ${POSTGRESQL_IMAGE}
     container_name: nexent-postgresql

deploy/docker/compose/docker-compose.yml

@@ -55,6 +55,22 @@ services:
         max-size: "100m" # Maximum size of a single log file
         max-file: "3" # Maximum number of log files to keep
 
+  nexent-qdrant:
+    image: ${QDRANT_IMAGE}
+    container_name: nexent-qdrant
+    volumes:
+      - ${ROOT_DIR}/qdrant:/qdrant/storage
+    ports:
+      - "6333:6333" # HTTP API
+    networks:
+      - nexent
+    restart: always
+    logging:
+      driver: "json-file"
+      options:
+        max-size: "100m"
+        max-file: "3"
+
   nexent-postgresql:
     image: ${POSTGRESQL_IMAGE}
     container_name: nexent-postgresql

deploy/docker/deploy.sh

@@ -1064,7 +1064,7 @@ deploy_infrastructure() {
   INFRA_SERVICES=""
 
   if deployment_csv_contains "$DEPLOYMENT_COMPONENTS" "infrastructure"; then
-    INFRA_SERVICES="nexent-elasticsearch nexent-postgresql nexent-minio redis"
+    INFRA_SERVICES="nexent-elasticsearch nexent-qdrant nexent-postgresql nexent-minio redis"
   fi
 
   # Add openssh-server if Terminal tool container is enabled

deploy/env/.env.example

@@ -30,9 +30,15 @@ TABLE_TRANSFORMER_MODEL_PATH=/opt/models/table-transformer-structure-recognition
 UNSTRUCTURED_DEFAULT_MODEL_INITIALIZE_PARAMS_JSON_PATH=/opt/models/yolox/config.json
 
 # Elasticsearch Service
+VECTOR_DATABASE_TYPE=elasticsearch
 ELASTICSEARCH_HOST=http://nexent-elasticsearch:9200
 ELASTIC_PASSWORD=nexent@2025
 
+# Qdrant Service
+QDRANT_URL=http://nexent-qdrant:6333
+QDRANT_API_KEY=
+QDRANT_TIMEOUT=20
+
 # Elasticsearch Memory Configuration
 ES_JAVA_OPTS="-Xms2g -Xmx2g"
 

deploy/env/image-source.general.env

@@ -4,6 +4,7 @@ NEXENT_DATA_PROCESS_IMAGE=nexent/nexent-data-process:${APP_VERSION}
 NEXENT_MCP_DOCKER_IMAGE=nexent/nexent-mcp:${APP_VERSION}
 
 ELASTICSEARCH_IMAGE=docker.elastic.co/elasticsearch/elasticsearch:8.17.4
+QDRANT_IMAGE=qdrant/qdrant:v1.18.2
 POSTGRESQL_IMAGE=postgres:15-alpine
 REDIS_IMAGE=redis:alpine
 MINIO_IMAGE=quay.io/minio/minio:RELEASE.2023-12-20T01-00-02Z

deploy/env/image-source.mainland.env

@@ -4,6 +4,7 @@ NEXENT_DATA_PROCESS_IMAGE=ccr.ccs.tencentyun.com/nexent-hub/nexent-data-process:
 NEXENT_MCP_DOCKER_IMAGE=ccr.ccs.tencentyun.com/nexent-hub/nexent-mcp:${APP_VERSION}
 
 ELASTICSEARCH_IMAGE=elastic.m.daocloud.io/elasticsearch/elasticsearch:8.17.4
+QDRANT_IMAGE=docker.m.daocloud.io/qdrant/qdrant:v1.18.2
 POSTGRESQL_IMAGE=docker.m.daocloud.io/postgres:15-alpine
 REDIS_IMAGE=docker.m.daocloud.io/redis:alpine
 MINIO_IMAGE=quay.m.daocloud.io/minio/minio:RELEASE.2023-12-20T01-00-02Z

deploy/k8s/helm/nexent/Chart.yaml

@@ -22,6 +22,10 @@ dependencies:
     version: 0.1.0
     repository: "file://./charts/nexent-elasticsearch"
     condition: nexent-elasticsearch.enabled
+  - name: nexent-qdrant
+    version: 0.1.0
+    repository: "file://./charts/nexent-qdrant"
+    condition: nexent-qdrant.enabled
   - name: nexent-postgresql
     version: 0.1.0
     repository: "file://./charts/nexent-postgresql"

deploy/k8s/helm/nexent/charts/nexent-common/templates/configmap.yaml

@@ -10,12 +10,15 @@ data:
   # Deployment version
   DEPLOYMENT_VERSION: {{ .Values.global.deploymentVersion | quote }}
 
-  # Elasticsearch Service
+  # Vector Database Services
+  VECTOR_DATABASE_TYPE: {{ .Values.config.vectorDatabase.type | quote }}
   ELASTICSEARCH_HOST: {{ .Values.config.elasticsearch.host | quote }}
   ES_JAVA_OPTS: {{ .Values.config.elasticsearch.javaOpts | quote }}
   ES_DISK_WATERMARK_LOW: {{ .Values.config.elasticsearch.diskWatermarkLow | quote }}
   ES_DISK_WATERMARK_HIGH: {{ .Values.config.elasticsearch.diskWatermarkHigh | quote }}
   ES_DISK_WATERMARK_FLOOD_STAGE: {{ .Values.config.elasticsearch.diskWatermarkFloodStage | quote }}
+  QDRANT_URL: {{ .Values.config.qdrant.url | quote }}
+  QDRANT_TIMEOUT: {{ .Values.config.qdrant.timeout | quote }}
 
   # Service URLs (internal)
   CONFIG_SERVICE_URL: {{ .Values.config.services.configUrl | quote }}

deploy/k8s/helm/nexent/charts/nexent-common/templates/secrets.yaml

@@ -11,6 +11,9 @@ data:
   {{- if .Values.secrets.elasticsearchApiKey }}
   ELASTICSEARCH_API_KEY: {{ .Values.secrets.elasticsearchApiKey | b64enc | quote }}
   {{- end }}
+  {{- if .Values.secrets.qdrantApiKey }}
+  QDRANT_API_KEY: {{ .Values.secrets.qdrantApiKey | b64enc | quote }}
+  {{- end }}
 
   NEXENT_POSTGRES_PASSWORD: {{ .Values.secrets.postgresPassword | b64enc | quote }}
 

deploy/k8s/helm/nexent/charts/nexent-common/values.yaml

@@ -54,12 +54,17 @@
     endpoint: "http://nexent-minio:9000"
     region: "cn-north-1"
     defaultBucket: "nexent"
+  vectorDatabase:
+    type: "elasticsearch"
   elasticsearch:
     host: "http://nexent-elasticsearch:9200"
     javaOpts: "-Xms2g -Xmx2g"
     diskWatermarkLow: "85%"
     diskWatermarkHigh: "90%"
     diskWatermarkFloodStage: "95%"
+  qdrant:
+    url: "http://nexent-qdrant:6333"
+    timeout: "20"
   skipProxy: "true"
   umask: "0022"
   isDeployedByKubernetes: "true"
@@ -189,6 +194,7 @@
 secrets:
   elasticPassword: "nexent@2025"
   elasticsearchApiKey: ""
+  qdrantApiKey: ""
   postgresPassword: "nexent@4321"
   minio:
     rootUser: "nexent"

deploy/k8s/helm/nexent/charts/nexent-qdrant/Chart.yaml

@@ -0,0 +1,10 @@
+apiVersion: v2
+name: nexent-qdrant
+description: Qdrant vector database for Nexent
+type: application
+version: 0.1.0
+appVersion: "v1.18.2"
+keywords:
+  - nexent
+  - qdrant
+  - vector-database

deploy/k8s/helm/nexent/charts/nexent-qdrant/templates/deployment.yaml

@@ -0,0 +1,53 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: nexent-qdrant
+  namespace: {{ .Values.global.namespace }}
+  labels:
+    app: nexent-qdrant
+  annotations:
+    "helm.sh/hook-weight": "-1"
+spec:
+  replicas: {{ .Values.replicaCount }}
+  selector:
+    matchLabels:
+      app: nexent-qdrant
+  template:
+    metadata:
+      labels:
+        app: nexent-qdrant
+    spec:
+      securityContext:
+        fsGroup: 1000
+      containers:
+        - name: qdrant
+          image: "{{ .Values.image.repository }}:{{ .Values.image.tag }}"
+          imagePullPolicy: {{ .Values.image.pullPolicy }}
+          ports:
+            - containerPort: 6333
+              name: http
+          volumeMounts:
+            - name: qdrant-data
+              mountPath: /qdrant/storage
+          resources:
+            requests:
+              memory: {{ .Values.resources.requests.memory }}
+              cpu: {{ .Values.resources.requests.cpu }}
+            limits:
+              memory: {{ .Values.resources.limits.memory }}
+              cpu: {{ .Values.resources.limits.cpu }}
+          livenessProbe:
+            tcpSocket:
+              port: 6333
+            initialDelaySeconds: 30
+            periodSeconds: 10
+            timeoutSeconds: 5
+          readinessProbe:
+            tcpSocket:
+              port: 6333
+            initialDelaySeconds: 10
+            periodSeconds: 5
+      volumes:
+        - name: qdrant-data
+          persistentVolumeClaim:
+            claimName: {{ default "nexent-qdrant" .Values.persistence.existingClaim }}

deploy/k8s/helm/nexent/charts/nexent-qdrant/templates/service.yaml

@@ -0,0 +1,18 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: nexent-qdrant
+  namespace: {{ .Values.global.namespace }}
+  labels:
+    app: nexent-qdrant
+spec:
+  type: {{ default "ClusterIP" .Values.service.type }}
+  ports:
+    - port: 6333
+      targetPort: 6333
+      name: http
+      {{- if and (eq (default "ClusterIP" .Values.service.type) "NodePort") .Values.service.nodePorts.http }}
+      nodePort: {{ .Values.service.nodePorts.http }}
+      {{- end }}
+  selector:
+    app: nexent-qdrant