Modifying DBCopilot to support managed Identity (#3075)

* Fix DBCopilot Vulnerabilities * fix * fix * fix * fix
Azure · Jun 24, 2024 · f2084d6 · f2084d6
1 parent 7aa989c
commit f2084d6
Show file tree

Hide file tree

Showing 14 changed files with 60 additions and 61 deletions.
diff --git a/assets/large_language_models/components_pipelines/data_ingestion_db_to_acs/spec.yaml b/assets/large_language_models/components_pipelines/data_ingestion_db_to_acs/spec.yaml
@@ -4,7 +4,7 @@ tags:
   Preview: ""
 name: llm_ingest_db_to_acs
 display_name: LLM - SQL Datastore to ACS Pipeline
-version: 0.0.87
+version: 0.0.88
 description: Single job pipeline to chunk data from AzureML sql data store, and create ACS embeddings index
 settings:
   default_compute: serverless
@@ -109,10 +109,11 @@ inputs:
     type: string
     optional: true
     description: "The instruct template for the LLM."
-  user_managed_identity_client_id:
-    type: string
+  managed_identity_enabled:
+    type: boolean
+    default: flase
     optional: true
-    description: "The user identity client_id for the deployment."
+    description: "Whether to connect using managed identity."
 outputs:
   grounding_index:
     type: uri_folder
@@ -152,8 +153,8 @@ jobs:
         type: uri_folder
       output_grounding_context_file: ${{parent.outputs.db_context}}
     environment_variables:
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
-    component: "azureml:llm_dbcopilot_grounding:0.0.61"
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
+    component: "azureml:llm_dbcopilot_grounding:0.0.62"
     type: command
   generate_meta_embeddings:
     type: command
@@ -220,7 +221,7 @@ jobs:
     #########################################
   db_sample_loading_generator:
     type: command
-    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.36"
+    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.37"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -301,7 +302,7 @@ jobs:
     environment_variables:
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_EMBEDDING: ${{parent.inputs.embedding_connection}}
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_CHAT: ${{parent.inputs.llm_connection}}
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -335,5 +336,5 @@ jobs:
         path: ${{parent.inputs.include_views}}
       instruct_template:
         path: ${{parent.inputs.instruct_template}}
-    component: "azureml:llm_dbcopilot_create_promptflow:0.0.61"
+    component: "azureml:llm_dbcopilot_create_promptflow:0.0.62"
     type: command
diff --git a/assets/large_language_models/components_pipelines/data_ingestion_db_to_faiss/spec.yaml b/assets/large_language_models/components_pipelines/data_ingestion_db_to_faiss/spec.yaml
@@ -4,7 +4,7 @@ tags:
   Preview: ""
 name: llm_ingest_db_to_faiss
 display_name: LLM - SQL Datastore to FAISS Pipeline
-version: 0.0.87
+version: 0.0.88
 description: Single job pipeline to chunk data from AzureML sql data store, and create FAISS embeddings index
 settings:
   default_compute: serverless
@@ -99,10 +99,11 @@ inputs:
     type: string
     optional: true
     description: "The instruct template for the LLM."
-  user_managed_identity_client_id:
-    type: string
+  managed_identity_enabled:
+    type: boolean
+    default: flase
     optional: true
-    description: "The user identity client_id for the deployment."
+    description: "Whether to connect using managed identity."
 outputs:
   grounding_index:
     type: uri_folder
@@ -142,8 +143,8 @@ jobs:
         type: uri_folder
       output_grounding_context_file: ${{parent.outputs.db_context}}
     environment_variables:
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
-    component: "azureml:llm_dbcopilot_grounding:0.0.61"
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
+    component: "azureml:llm_dbcopilot_grounding:0.0.62"
     type: command
   generate_meta_embeddings:
     type: command
@@ -208,7 +209,7 @@ jobs:
   #########################################
   db_sample_loading_generator:
     type: command
-    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.36"
+    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.37"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -285,7 +286,7 @@ jobs:
     environment_variables:
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_EMBEDDING: ${{parent.inputs.embedding_connection}}
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_CHAT: ${{parent.inputs.llm_connection}}
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -319,5 +320,5 @@ jobs:
         path: ${{parent.inputs.include_views}}
       instruct_template:
         path: ${{parent.inputs.instruct_template}}
-    component: "azureml:llm_dbcopilot_create_promptflow:0.0.61"
+    component: "azureml:llm_dbcopilot_create_promptflow:0.0.62"
     type: command
diff --git a/assets/large_language_models/components_pipelines/data_ingestion_dbcopilot_acs_e2e/spec.yaml b/assets/large_language_models/components_pipelines/data_ingestion_dbcopilot_acs_e2e/spec.yaml
@@ -2,7 +2,7 @@ $schema: https://azuremlschemas.azureedge.net/latest/pipelineComponent.schema.js
 type: pipeline
 
 name: llm_ingest_dbcopilot_acs_e2e
-version: 0.0.55
+version: 0.0.56
 display_name: Data Ingestion for DB Data Output to ACS E2E Deployment
 description: Single job pipeline to chunk data from AzureML DB Datastore and create acs embeddings index
 
@@ -122,10 +122,11 @@ inputs:
     type: string
     optional: true
     description: "The instruct template for the LLM."
-  user_managed_identity_client_id:
-    type: string
+  managed_identity_enabled:
+    type: boolean
+    default: flase
     optional: true
-    description: "The user identity client_id for the deployment."
+    description: "Whether to connect using managed identity."
 outputs:
   grounding_index:
     type: uri_folder
@@ -135,7 +136,7 @@ jobs:
   #########################################
   db_meta_loading_generator:
     type: command
-    component: "azureml:llm_dbcopilot_grounding:0.0.61"
+    component: "azureml:llm_dbcopilot_grounding:0.0.62"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -160,7 +161,7 @@ jobs:
         type: uri_folder
         path: ${{parent.outputs.db_context}}
     environment_variables:
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
   #########################################
   generate_meta_embeddings:
     type: command
@@ -207,7 +208,7 @@ jobs:
   #########################################
   db_sample_loading_generator:
     type: command
-    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.36"
+    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.37"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -269,7 +270,7 @@ jobs:
   #########################################
   endpoint_deployment_job:
     type: command
-    component: "azureml:llm_dbcopilot_deploy_endpoint:0.0.36"
+    component: "azureml:llm_dbcopilot_deploy_endpoint:0.0.37"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -309,4 +310,4 @@ jobs:
     environment_variables:
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_EMBEDDING: ${{parent.inputs.embedding_connection}}
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_CHAT: ${{parent.inputs.llm_connection}}
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
diff --git a/...s/large_language_models/components_pipelines/data_ingestion_dbcopilot_faiss_e2e/spec.yaml b/...s/large_language_models/components_pipelines/data_ingestion_dbcopilot_faiss_e2e/spec.yaml
@@ -2,7 +2,7 @@ $schema: https://azuremlschemas.azureedge.net/latest/pipelineComponent.schema.js
 type: pipeline
 
 name: llm_ingest_dbcopilot_faiss_e2e
-version: 0.0.55
+version: 0.0.56
 display_name: Data Ingestion for DB Data Output to FAISS E2E Deployment
 description: Single job pipeline to chunk data from AzureML DB Datastore and create faiss embeddings index
 
@@ -112,10 +112,11 @@ inputs:
     type: string
     optional: true
     description: "The instruct template for the LLM."
-  user_managed_identity_client_id:
-    type: string
+  managed_identity_enabled:
+    type: boolean
+    default: flase
     optional: true
-    description: "The user identity client_id for the deployment."
+    description: "Whether to connect using managed identity."
 outputs:
   grounding_index:
     type: uri_folder
@@ -125,7 +126,7 @@ jobs:
   #########################################
   db_meta_loading_generator:
     type: command
-    component: "azureml:llm_dbcopilot_grounding:0.0.61"
+    component: "azureml:llm_dbcopilot_grounding:0.0.62"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -150,7 +151,7 @@ jobs:
         type: uri_folder
         path: ${{parent.outputs.db_context}}
     environment_variables:
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
   #########################################
   generate_meta_embeddings:
     type: command
@@ -195,7 +196,7 @@ jobs:
   #########################################
   db_sample_loading_generator:
     type: command
-    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.36"
+    component: "azureml:llm_dbcopilot_grounding_ground_samples:0.0.37"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -253,7 +254,7 @@ jobs:
   #########################################
   endpoint_deployment_job:
     type: command
-    component: "azureml:llm_dbcopilot_deploy_endpoint:0.0.36"
+    component: "azureml:llm_dbcopilot_deploy_endpoint:0.0.37"
     resources:
       instance_count: ${{parent.inputs.serverless_instance_count}}
       instance_type: ${{parent.inputs.serverless_instance_type}}
@@ -293,4 +294,4 @@ jobs:
     environment_variables:
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_EMBEDDING: ${{parent.inputs.embedding_connection}}
       AZUREML_WORKSPACE_CONNECTION_ID_AOAI_CHAT: ${{parent.inputs.llm_connection}}
-      USER_MANAGED_IDENTITY_CLIENT_ID: ${{parent.inputs.user_managed_identity_client_id}}
+      MANAGED_IDENTITY_ENABLED: ${{parent.inputs.managed_identity_enabled}}
diff --git a/assets/large_language_models/dbcopilot/components/llm_dbcopilot_create_promptflow/spec.yaml b/assets/large_language_models/dbcopilot/components/llm_dbcopilot_create_promptflow/spec.yaml
@@ -4,7 +4,7 @@ tags:
   Preview: ""
 name: llm_dbcopilot_create_promptflow
 display_name: LLM - Create DBCopilot Prompt Flow
-version: 0.0.61
+version: 0.0.62
 inputs:
   index_name:
     type: string

diff --git a/assets/large_language_models/dbcopilot/components/llm_dbcopilot_deploy_endpoint/spec.yaml b/assets/large_language_models/dbcopilot/components/llm_dbcopilot_deploy_endpoint/spec.yaml
@@ -3,7 +3,7 @@ type: command
 tags: {}
 name: llm_dbcopilot_deploy_endpoint
 display_name: LLM - DBCopilot Deploy Endpoint Component
-version: 0.0.36
+version: 0.0.37
 inputs:
   deployment_name:
     type: string

diff --git a/assets/large_language_models/dbcopilot/components/llm_dbcopilot_grounding/spec.yaml b/assets/large_language_models/dbcopilot/components/llm_dbcopilot_grounding/spec.yaml
@@ -3,7 +3,7 @@ type: command
 tags:
   Preview: ""
 name: llm_dbcopilot_grounding
-version: 0.0.61
+version: 0.0.62
 inputs:
   asset_uri:
     type: string

diff --git a/...rge_language_models/dbcopilot/components/llm_dbcopilot_grounding_ground_samples/spec.yaml b/...rge_language_models/dbcopilot/components/llm_dbcopilot_grounding_ground_samples/spec.yaml
@@ -3,7 +3,7 @@ type: command
 tags: {}
 name: llm_dbcopilot_grounding_ground_samples
 display_name: LLM - DBCopilot Grounding Ground Samples Component
-version: 0.0.36
+version: 0.0.37
 inputs:
   grounding_context:
     type: uri_folder

diff --git a/assets/large_language_models/dbcopilot/components/src/endpoint_deployment.py b/assets/large_language_models/dbcopilot/components/src/endpoint_deployment.py
@@ -104,7 +104,7 @@ def deploy(
             logging.info("dumped secrets to secrets.json")
             with open(os.path.join(code_dir, "configs.json"), "w") as f:
                 json.dump([asdict(config)], f)
-            user_managed_identity_client_id = os.getenv("USER_MANAGED_IDENTITY_CLIENT_ID", None)
+            managed_identity_enabled = os.getenv("MANAGED_IDENTITY_ENABLED", None)
             self._deploy_endpoint(
                 mir_environment,
                 endpoint_name,
@@ -113,7 +113,7 @@ def deploy(
                 score_script="score_zero.py",
                 extra_environment_variables={
                     "INSTRUCT_TEMPLATE": instruct_template,
-                    "USER_MANAGED_IDENTITY_CLIENT_ID": user_managed_identity_client_id,
+                    "MANAGED_IDENTITY_ENABLED": managed_identity_enabled,
                 },
                 sku=sku,
             )
diff --git a/assets/large_language_models/dbcopilot/components/src/prompt_flow_creation.py b/assets/large_language_models/dbcopilot/components/src/prompt_flow_creation.py
@@ -59,7 +59,7 @@ def create(
         datastore_uri = get_datastore_uri(workspace, asset_uri)
         logging.info(f"Datastore uri: {datastore_uri}")
 
-        user_managed_identity_client_id = os.environ.get("USER_MANAGED_IDENTITY_CLIENT_ID", None)
+        managed_identity_enabled = os.environ.get("MANAGED_IDENTITY_ENABLED", None)
         embedding_connection_id = os.environ.get(
             "AZUREML_WORKSPACE_CONNECTION_ID_AOAI_EMBEDDING", None
         )
@@ -127,7 +127,7 @@ def create(
             "AZUREML_WORKSPACE_NAME": self.workspace.name,
             "AZUREML_SUBSCRIPTION_ID": self.workspace.subscription_id,
             "AZUREML_RESOURCE_GROUP": self.workspace.resource_group,
-            "USER_MANAGED_IDENTITY_CLIENT_ID": user_managed_identity_client_id,
+            "MANAGED_IDENTITY_ENABLED": managed_identity_enabled,
         }
         base_run = pf_client.run(
             flow=flow,

diff --git a/assets/large_language_models/dbcopilot/environments/dbcopilot_embeddings/context/Dockerfile b/assets/large_language_models/dbcopilot/environments/dbcopilot_embeddings/context/Dockerfile
@@ -8,8 +8,12 @@ RUN apt-get update && apt-get install -y \
     curl \
     gnupg \
     unixodbc-dev \
-    git=1:2.25.1-1ubuntu3.12\
-    git-man=1:2.25.1-1ubuntu3.12 &&\
+    libc-bin=2.31-0ubuntu9.16 \
+    libc6-dev=2.31-0ubuntu9.16 \
+    libc6=2.31-0ubuntu9.16 \
+    libc-dev-bin=2.31-0ubuntu9.16 \
+    git \
+    git-man &&\
     rm -rf /var/lib/apt/lists/*
 
 # Install MS SQL ODBC Driver
@@ -35,8 +39,8 @@ RUN /bin/bash -c "source /opt/miniconda/etc/profile.d/conda.sh && \
     conda activate $AZUREML_CONDA_ENVIRONMENT_PATH && \
     pip install --upgrade pip && \
     pip install  -r requirements.txt && \
-    pip install  promptflow-vectordb==0.2.9 && \
-    pip install  https://ragsample.blob.core.windows.net/ragdata/wheels/dbcopilot/db_copilot_tool-0.1.22-py3-none-any.whl &&\
+    pip install  promptflow-vectordb==0.2.10 && \
+    pip install  https://ragsample.blob.core.windows.net/ragdata/wheels/dbcopilot/db_copilot_tool-0.1.23-py3-none-any.whl &&\
     pip install  cryptography==42.0.5 langchain==0.1.11 idna==3.7 sqlparse==0.5.0 gunicorn==22.0.0 Werkzeug==3.0.3 requests==2.32.0"
 
 # Fix vunerabilities

diff --git a/...arge_language_models/dbcopilot/environments/dbcopilot_embeddings/context/requirements.txt b/...arge_language_models/dbcopilot/environments/dbcopilot_embeddings/context/requirements.txt
@@ -1,4 +1,4 @@
-azureml-rag[cognitive_search,data_generation]==0.2.29.1
+azureml-rag[cognitive_search,data_generation]==0.2.34
 azureml-contrib-services
 azure-identity==1.14.0
 azureml-core~=1.53.0

diff --git a/assets/large_language_models/dbcopilot/environments/dbcopilot_mir/context/Dockerfile b/assets/large_language_models/dbcopilot/environments/dbcopilot_mir/context/Dockerfile
@@ -9,15 +9,6 @@ RUN apt-get update \
     && curl https://packages.microsoft.com/config/ubuntu/22.04/prod.list | tee /etc/apt/sources.list.d/mssql-release.list \
     && apt-get update \
     && ACCEPT_EULA=Y apt-get install -y msodbcsql18=18.3.3.1-1 \
-    && apt-get install -y libpam0g=1.4.0-11ubuntu2.4 \
-    && apt-get install -y libexpat1=2.4.7-1ubuntu0.3 \
-    && apt-get install -y bash=5.1-6ubuntu1.1 \
-    && apt-get install -y libuuid1=2.37.2-4ubuntu3.4 \
-    && apt-get install -y libblkid1=2.37.2-4ubuntu3.4\
-    && apt-get install -y util-linux=2.37.2-4ubuntu3.4 \
-    && apt-get install -y mount=2.37.2-4ubuntu3.4 \
-    && apt-get install -y libsmartcols1=2.37.2-4ubuntu3.4 \
-    && apt-get install -y libmount1=2.37.2-4ubuntu3.4 \
     && apt-get clean \
     && rm -rf /var/lib/apt/lists/*
 
@@ -34,9 +25,9 @@ RUN python -m pip install --upgrade pip && \
     pip install -r requirements.txt && \
     # Install promptflow environment
     pip install promptflow-image-bundle[azure]==0.3.0 && \
-    pip install promptflow-vectordb==0.2.9 && \
+    pip install promptflow-vectordb==0.2.10 && \
     ## Install dbcopilot
-    pip install https://ragsample.blob.core.windows.net/ragdata/wheels/dbcopilot/db_copilot_tool-0.1.22-py3-none-any.whl && \
+    pip install https://ragsample.blob.core.windows.net/ragdata/wheels/dbcopilot/db_copilot_tool-0.1.23-py3-none-any.whl && \
     ## Fix vulnerabilities
     pip install idna==3.7
 

diff --git a/assets/large_language_models/dbcopilot/environments/dbcopilot_mir/context/requirements.txt b/assets/large_language_models/dbcopilot/environments/dbcopilot_mir/context/requirements.txt
@@ -1,4 +1,4 @@
-azureml-rag[cognitive_search,data_generation]==0.2.29.1
+azureml-rag[cognitive_search,data_generation]==0.2.34
 plotly~=5.13.1
 azure-kusto-data==4.2.0
 recognizers-text-suite~=1.0.2a2