Refactor environment variable handling using ConfigMap for Kubernetes deployment (#515)

---------

Co-authored-by: Reese Jenner <reesevader@hotmail.co.uk>
Co-authored-by: Yuhong Sun <yuhongsun96@gmail.com>

backend/danswer/configs/app_configs.py

@@ -44,7 +44,7 @@ MASK_CREDENTIAL_PREFIX = (
 
 SECRET = os.environ.get("SECRET", "")
 SESSION_EXPIRE_TIME_SECONDS = int(
-    os.environ.get("SESSION_EXPIRE_TIME_SECONDS", 86400)
+    os.environ.get("SESSION_EXPIRE_TIME_SECONDS") or 86400
 )  # 1 day
 
 # set `VALID_EMAIL_DOMAINS` to a comma seperated list of domains in order to
@@ -189,22 +189,11 @@ MODEL_SERVER_HOST = os.environ.get("MODEL_SERVER_HOST") or None
 MODEL_SERVER_ALLOWED_HOST = os.environ.get("MODEL_SERVER_HOST") or "0.0.0.0"
 MODEL_SERVER_PORT = int(os.environ.get("MODEL_SERVER_PORT") or "9000")
 
-EMBEDDING_MODEL_SERVER_HOST = (
-    os.environ.get("EMBEDDING_MODEL_SERVER_HOST") or MODEL_SERVER_HOST
-)
-CROSS_ENCODER_MODEL_SERVER_HOST = (
-    os.environ.get("CROSS_ENCODER_MODEL_SERVER_HOST") or MODEL_SERVER_HOST
-)
-INTENT_MODEL_SERVER_HOST = (
-    os.environ.get("INTENT_MODEL_SERVER_HOST") or MODEL_SERVER_HOST
-)
-
 # specify this env variable directly to have a different model server for the background
 # indexing job vs the api server so that background indexing does not effect query-time
 # performance
-BACKGROUND_JOB_EMBEDDING_MODEL_SERVER_HOST = (
+INDEXING_MODEL_SERVER_HOST = (
-    os.environ.get("BACKGROUND_JOB_EMBEDDING_MODEL_SERVER_HOST")
+    os.environ.get("INDEXING_MODEL_SERVER_HOST") or MODEL_SERVER_HOST
-    or EMBEDDING_MODEL_SERVER_HOST
 )
 
 

backend/danswer/search/search_nlp_models.py

@@ -9,11 +9,9 @@ from sentence_transformers import SentenceTransformer  # type: ignore
 from transformers import AutoTokenizer  # type: ignore
 from transformers import TFDistilBertForSequenceClassification  # type: ignore
 
-from danswer.configs.app_configs import BACKGROUND_JOB_EMBEDDING_MODEL_SERVER_HOST
-from danswer.configs.app_configs import CROSS_ENCODER_MODEL_SERVER_HOST
 from danswer.configs.app_configs import CURRENT_PROCESS_IS_AN_INDEXING_JOB
-from danswer.configs.app_configs import EMBEDDING_MODEL_SERVER_HOST
+from danswer.configs.app_configs import INDEXING_MODEL_SERVER_HOST
-from danswer.configs.app_configs import INTENT_MODEL_SERVER_HOST
+from danswer.configs.app_configs import MODEL_SERVER_HOST
 from danswer.configs.app_configs import MODEL_SERVER_PORT
 from danswer.configs.model_configs import CROSS_EMBED_CONTEXT_SIZE
 from danswer.configs.model_configs import CROSS_ENCODER_MODEL_ENSEMBLE
@@ -99,12 +97,13 @@ def get_local_intent_model(
 
 
 def build_model_server_url(
-    model_server_host: str,
+    model_server_host: str | None,
     model_server_port: int | None,
-) -> str:
+) -> str | None:
-    model_server_url = model_server_host + (
+    if not model_server_host or model_server_port is None:
-        f":{model_server_port}" if model_server_port else ""
+        return None
-    )
+
+    model_server_url = f"{model_server_host}:{model_server_port}"
 
     # use protocol if provided
     if "http" in model_server_url:
@@ -119,27 +118,23 @@ class EmbeddingModel:
         self,
         model_name: str = DOCUMENT_ENCODER_MODEL,
         max_seq_length: int = DOC_EMBEDDING_CONTEXT_SIZE,
-        # `model_server_host` one has to default to `None` since it's
+        model_server_host: str | None = MODEL_SERVER_HOST,
-        # default value is conditional
+        indexing_model_server_host: str | None = INDEXING_MODEL_SERVER_HOST,
-        model_server_host: str | None = None,
         model_server_port: int = MODEL_SERVER_PORT,
+        is_indexing: bool = CURRENT_PROCESS_IS_AN_INDEXING_JOB,
     ) -> None:
-        if model_server_host is None:
-            model_server_host = (
-                BACKGROUND_JOB_EMBEDDING_MODEL_SERVER_HOST
-                if CURRENT_PROCESS_IS_AN_INDEXING_JOB
-                else EMBEDDING_MODEL_SERVER_HOST
-            )
-
         self.model_name = model_name
         self.max_seq_length = max_seq_length
+
+        used_model_server_host = (
+            indexing_model_server_host if is_indexing else model_server_host
+        )
+
+        model_server_url = build_model_server_url(
+            used_model_server_host, model_server_port
+        )
         self.embed_server_endpoint = (
-            (
+            model_server_url + "/encoder/bi-encoder-embed" if model_server_url else None
-                build_model_server_url(model_server_host, model_server_port)
-                + "/encoder/bi-encoder-embed"
-            )
-            if model_server_host
-            else None
         )
 
     def load_model(self) -> SentenceTransformer | None:
@@ -182,17 +177,16 @@ class CrossEncoderEnsembleModel:
         self,
         model_names: list[str] = CROSS_ENCODER_MODEL_ENSEMBLE,
         max_seq_length: int = CROSS_EMBED_CONTEXT_SIZE,
-        model_server_host: str | None = CROSS_ENCODER_MODEL_SERVER_HOST,
+        model_server_host: str | None = MODEL_SERVER_HOST,
         model_server_port: int = MODEL_SERVER_PORT,
     ) -> None:
         self.model_names = model_names
         self.max_seq_length = max_seq_length
+
+        model_server_url = build_model_server_url(model_server_host, model_server_port)
         self.rerank_server_endpoint = (
-            (
+            model_server_url + "/encoder/cross-encoder-scores"
-                build_model_server_url(model_server_host, model_server_port)
+            if model_server_url
-                + "/encoder/cross-encoder-scores"
-            )
-            if model_server_host
             else None
         )
 
@@ -237,18 +231,15 @@ class IntentModel:
         self,
         model_name: str = INTENT_MODEL_VERSION,
         max_seq_length: int = QUERY_MAX_CONTEXT_SIZE,
-        model_server_host: str | None = INTENT_MODEL_SERVER_HOST,
+        model_server_host: str | None = MODEL_SERVER_HOST,
         model_server_port: int = MODEL_SERVER_PORT,
     ) -> None:
         self.model_name = model_name
         self.max_seq_length = max_seq_length
+
+        model_server_url = build_model_server_url(model_server_host, model_server_port)
         self.intent_server_endpoint = (
-            (
+            model_server_url + "/custom/intent-model" if model_server_url else None
-                build_model_server_url(model_server_host, model_server_port)
-                + "/custom/intent-model"
-            )
-            if model_server_host
-            else None
         )
 
     def load_model(self) -> SentenceTransformer | None:

deployment/docker_compose/docker-compose.dev.yml

@@ -16,6 +16,13 @@ services:
     ports:
       - "8080:8080"
     environment:
+      # Auth Settings
+      - AUTH_TYPE=${AUTH_TYPE:-disabled}
+      - SESSION_EXPIRE_TIME_SECONDS=${SESSION_EXPIRE_TIME_SECONDS:-}
+      - VALID_EMAIL_DOMAINS=${VALID_EMAIL_DOMAINS:-}
+      - GOOGLE_OAUTH_CLIENT_ID=${GOOGLE_OAUTH_CLIENT_ID:-}
+      - GOOGLE_OAUTH_CLIENT_SECRET=${GOOGLE_OAUTH_CLIENT_SECRET:-}
+      # Gen AI Settings
       - GEN_AI_MODEL_PROVIDER=${GEN_AI_MODEL_PROVIDER:-openai}
       - GEN_AI_MODEL_VERSION=${GEN_AI_MODEL_VERSION:-gpt-3.5-turbo}
       - FAST_GEN_AI_MODEL_VERSION=${FAST_GEN_AI_MODEL_VERSION:-gpt-3.5-turbo}
@@ -23,43 +30,34 @@ services:
       - GEN_AI_API_ENDPOINT=${GEN_AI_API_ENDPOINT:-}
       - GEN_AI_API_VERSION=${GEN_AI_API_VERSION:-}
       - GEN_AI_LLM_PROVIDER_TYPE=${GEN_AI_LLM_PROVIDER_TYPE:-}
-      - NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL=${NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL:-}
-      - POSTGRES_HOST=relational_db
-      - VESPA_HOST=index
-      - AUTH_TYPE=${AUTH_TYPE:-disabled}
       - QA_TIMEOUT=${QA_TIMEOUT:-}
-      - VALID_EMAIL_DOMAINS=${VALID_EMAIL_DOMAINS:-}
+      - NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL=${NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL:-}
-      - GOOGLE_OAUTH_CLIENT_ID=${GOOGLE_OAUTH_CLIENT_ID:-}
-      - GOOGLE_OAUTH_CLIENT_SECRET=${GOOGLE_OAUTH_CLIENT_SECRET:-}
-      - NOTION_CONNECTOR_ENABLE_RECURSIVE_PAGE_LOOKUP=${NOTION_CONNECTOR_ENABLE_RECURSIVE_PAGE_LOOKUP:-}
       - DISABLE_LLM_FILTER_EXTRACTION=${DISABLE_LLM_FILTER_EXTRACTION:-}
       - DISABLE_LLM_CHUNK_FILTER=${DISABLE_LLM_CHUNK_FILTER:-}
       - DISABLE_LLM_CHOOSE_SEARCH=${DISABLE_LLM_CHOOSE_SEARCH:-}
-      # Recency Bias for search results, decay at 1 / (1 + DOC_TIME_DECAY * x years)
+      # Query Options
-      - DOC_TIME_DECAY=${DOC_TIME_DECAY:-}
+      - DOC_TIME_DECAY=${DOC_TIME_DECAY:-}  # Recency Bias for search results, decay at 1 / (1 + DOC_TIME_DECAY * x years)
-      # Hybrid Search Alpha (0 for entirely keyword, 1 for entirely vector)
+      - HYBRID_ALPHA=${HYBRID_ALPHA:-}  # Hybrid Search Alpha (0 for entirely keyword, 1 for entirely vector)
-      - HYBRID_ALPHA=${HYBRID_ALPHA:-}
+      - EDIT_KEYWORD_QUERY=${EDIT_KEYWORD_QUERY:-}
+      - MULTILINGUAL_QUERY_EXPANSION=${MULTILINGUAL_QUERY_EXPANSION:-}
+      - QA_PROMPT_OVERRIDE=${QA_PROMPT_OVERRIDE:-}
+      # Other services
+      - POSTGRES_HOST=relational_db
+      - VESPA_HOST=index
+      - WEB_DOMAIN=${WEB_DOMAIN:-}  # For frontend redirect auth purpose
       # Don't change the NLP model configs unless you know what you're doing
       - DOCUMENT_ENCODER_MODEL=${DOCUMENT_ENCODER_MODEL:-}
       - NORMALIZE_EMBEDDINGS=${NORMALIZE_EMBEDDINGS:-}
       - ASYM_QUERY_PREFIX=${ASYM_QUERY_PREFIX:-}
-      - ASYM_PASSAGE_PREFIX=${ASYM_PASSAGE_PREFIX:-}
       - ENABLE_RERANKING_REAL_TIME_FLOW=${ENABLE_RERANKING_REAL_TIME_FLOW:-}
-      - QA_PROMPT_OVERRIDE=${QA_PROMPT_OVERRIDE:-}
+      - ENABLE_RERANKING_ASYNC_FLOW=${ENABLE_RERANKING_ASYNC_FLOW:-}
-      - EDIT_KEYWORD_QUERY=${EDIT_KEYWORD_QUERY:-}
-      - MULTILINGUAL_QUERY_EXPANSION=${MULTILINGUAL_QUERY_EXPANSION:-}
       - MODEL_SERVER_HOST=${MODEL_SERVER_HOST:-}
-      - EMBEDDING_MODEL_SERVER_HOST=${EMBEDDING_MODEL_SERVER_HOST:-}
-      - CROSS_ENCODER_MODEL_SERVER_HOST=${CROSS_ENCODER_MODEL_SERVER_HOST:-}
-      - INTENT_MODEL_SERVER_HOST=${INTENT_MODEL_SERVER_HOST:-}
       - MODEL_SERVER_PORT=${MODEL_SERVER_PORT:-}
       # Leave this on pretty please? Nothing sensitive is collected!
       # https://docs.danswer.dev/more/telemetry
       - DISABLE_TELEMETRY=${DISABLE_TELEMETRY:-}
-      # Set to debug to get more fine-grained logs
+      - LOG_LEVEL=${LOG_LEVEL:-info}  # Set to debug to get more fine-grained logs
-      - LOG_LEVEL=${LOG_LEVEL:-info}
+      - LOG_ALL_MODEL_INTERACTIONS=${LOG_ALL_MODEL_INTERACTIONS:-}  # Log all of the prompts to the LLM
-      # Log all of the prompts to the LLM
-      - LOG_ALL_MODEL_INTERACTIONS=${LOG_ALL_MODEL_INTERACTIONS:-}
       # If set to `true` will enable additional logs about Vespa query performance
       # (time spent on finding the right docs + time spent fetching summaries from disk)
       - LOG_VESPA_TIMING_INFORMATION=${LOG_VESPA_TIMING_INFORMATION:-}
@@ -80,6 +78,7 @@ services:
       - index
     restart: always
     environment:
+      # Gen AI Settings (Needed by DanswerBot)
       - GEN_AI_MODEL_PROVIDER=${GEN_AI_MODEL_PROVIDER:-openai}
       - GEN_AI_MODEL_VERSION=${GEN_AI_MODEL_VERSION:-gpt-3.5-turbo}
       - FAST_GEN_AI_MODEL_VERSION=${FAST_GEN_AI_MODEL_VERSION:-gpt-3.5-turbo}
@@ -87,19 +86,38 @@ services:
       - GEN_AI_API_ENDPOINT=${GEN_AI_API_ENDPOINT:-}
       - GEN_AI_API_VERSION=${GEN_AI_API_VERSION:-}
       - GEN_AI_LLM_PROVIDER_TYPE=${GEN_AI_LLM_PROVIDER_TYPE:-}
+      - QA_TIMEOUT=${QA_TIMEOUT:-}
+      - NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL=${NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL:-}
       - DISABLE_LLM_FILTER_EXTRACTION=${DISABLE_LLM_FILTER_EXTRACTION:-}
       - DISABLE_LLM_CHUNK_FILTER=${DISABLE_LLM_CHUNK_FILTER:-}
       - DISABLE_LLM_CHOOSE_SEARCH=${DISABLE_LLM_CHOOSE_SEARCH:-}
+      # Query Options
+      - DOC_TIME_DECAY=${DOC_TIME_DECAY:-}  # Recency Bias for search results, decay at 1 / (1 + DOC_TIME_DECAY * x years)
+      - HYBRID_ALPHA=${HYBRID_ALPHA:-}  # Hybrid Search Alpha (0 for entirely keyword, 1 for entirely vector)
+      - EDIT_KEYWORD_QUERY=${EDIT_KEYWORD_QUERY:-}
+      - MULTILINGUAL_QUERY_EXPANSION=${MULTILINGUAL_QUERY_EXPANSION:-}
+      - QA_PROMPT_OVERRIDE=${QA_PROMPT_OVERRIDE:-}
+      # Other Services
       - POSTGRES_HOST=relational_db
       - VESPA_HOST=index
+      - WEB_DOMAIN=${WEB_DOMAIN:-}  # For frontend redirect auth purpose for OAuth2 connectors
+      # Don't change the NLP model configs unless you know what you're doing
+      - DOCUMENT_ENCODER_MODEL=${DOCUMENT_ENCODER_MODEL:-}
+      - NORMALIZE_EMBEDDINGS=${NORMALIZE_EMBEDDINGS:-}
+      - ASYM_QUERY_PREFIX=${ASYM_QUERY_PREFIX:-}  # Needed by DanswerBot
+      - ASYM_PASSAGE_PREFIX=${ASYM_PASSAGE_PREFIX:-}
+      - MODEL_SERVER_HOST=${MODEL_SERVER_HOST:-}
+      - MODEL_SERVER_PORT=${MODEL_SERVER_PORT:-}
+      - INDEXING_MODEL_SERVER_HOST=${INDEXING_MODEL_SERVER_HOST:-}
+      - MIN_THREADS_ML_MODELS=${MIN_THREADS_ML_MODELS:-}
+      # Indexing Configs
       - NUM_INDEXING_WORKERS=${NUM_INDEXING_WORKERS:-}
       - DASK_JOB_CLIENT_ENABLED=${DASK_JOB_CLIENT_ENABLED:-}
-      # Connector Configs
       - CONTINUE_ON_CONNECTOR_FAILURE=${CONTINUE_ON_CONNECTOR_FAILURE:-}
+      - EXPERIMENTAL_CHECKPOINTING_ENABLED=${EXPERIMENTAL_CHECKPOINTING_ENABLED:-}
       - CONFLUENCE_CONNECTOR_LABELS_TO_SKIP=${CONFLUENCE_CONNECTOR_LABELS_TO_SKIP:-}
       - GONG_CONNECTOR_START_TIME=${GONG_CONNECTOR_START_TIME:-}
-      - EXPERIMENTAL_SIMPLE_JOB_CLIENT_ENABLED=${EXPERIMENTAL_SIMPLE_JOB_CLIENT_ENABLED:-}
+      - NOTION_CONNECTOR_ENABLE_RECURSIVE_PAGE_LOOKUP=${NOTION_CONNECTOR_ENABLE_RECURSIVE_PAGE_LOOKUP:-}
-      - EXPERIMENTAL_CHECKPOINTING_ENABLED=${EXPERIMENTAL_CHECKPOINTING_ENABLED:-}
       # Danswer SlackBot Configs
       - DANSWER_BOT_SLACK_APP_TOKEN=${DANSWER_BOT_SLACK_APP_TOKEN:-}
       - DANSWER_BOT_SLACK_BOT_TOKEN=${DANSWER_BOT_SLACK_BOT_TOKEN:-}
@@ -108,33 +126,13 @@ services:
       - DANSWER_BOT_RESPOND_EVERY_CHANNEL=${DANSWER_BOT_RESPOND_EVERY_CHANNEL:-}
       - DANSWER_BOT_DISABLE_COT=${DANSWER_BOT_DISABLE_COT:-}  # Currently unused
       - NOTIFY_SLACKBOT_NO_ANSWER=${NOTIFY_SLACKBOT_NO_ANSWER:-}
-      # Recency Bias for search results, decay at 1 / (1 + DOC_TIME_DECAY * x years)
+      # Logging
-      - DOC_TIME_DECAY=${DOC_TIME_DECAY:-}
-      # Hybrid Search Alpha (0 for entirely keyword, 1 for entirely vector)
-      - HYBRID_ALPHA=${HYBRID_ALPHA:-}
-      # Don't change the NLP model configs unless you know what you're doing
-      - DOCUMENT_ENCODER_MODEL=${DOCUMENT_ENCODER_MODEL:-}
-      - NORMALIZE_EMBEDDINGS=${NORMALIZE_EMBEDDINGS:-}
-      - SIM_SCORE_RANGE_LOW=${SIM_SCORE_RANGE_LOW:-}
-      - SIM_SCORE_RANGE_HIGH=${SIM_SCORE_RANGE_HIGH:-}
-      - ASYM_QUERY_PREFIX=${ASYM_QUERY_PREFIX:-}
-      - ASYM_PASSAGE_PREFIX=${ASYM_PASSAGE_PREFIX:-}
-      - ENABLE_RERANKING_ASYNC_FLOW=${ENABLE_RERANKING_ASYNC_FLOW:-}
-      - QA_PROMPT_OVERRIDE=${QA_PROMPT_OVERRIDE:-}
-      - EDIT_KEYWORD_QUERY=${EDIT_KEYWORD_QUERY:-}
-      - MULTILINGUAL_QUERY_EXPANSION=${MULTILINGUAL_QUERY_EXPANSION:-}
-      - MIN_THREADS_ML_MODELS=${MIN_THREADS_ML_MODELS:-}
-      - MODEL_SERVER_HOST=${MODEL_SERVER_HOST:-}
-      - EMBEDDING_MODEL_SERVER_HOST=${EMBEDDING_MODEL_SERVER_HOST:-}
-      - BACKGROUND_JOB_EMBEDDING_MODEL_SERVER_HOST=${BACKGROUND_JOB_EMBEDDING_MODEL_SERVER_HOST:-}
-      - MODEL_SERVER_PORT=${MODEL_SERVER_PORT:-}
       # Leave this on pretty please? Nothing sensitive is collected!
       # https://docs.danswer.dev/more/telemetry
       - DISABLE_TELEMETRY=${DISABLE_TELEMETRY:-}
-      # Set to debug to get more fine-grained logs
+      - LOG_LEVEL=${LOG_LEVEL:-info}  # Set to debug to get more fine-grained logs
-      - LOG_LEVEL=${LOG_LEVEL:-info}
+      - LOG_ALL_MODEL_INTERACTIONS=${LOG_ALL_MODEL_INTERACTIONS:-}  # Log all of the prompts to the LLM
-      # Log all of the prompts to the LLM
+      - LOG_VESPA_TIMING_INFORMATION=${LOG_VESPA_TIMING_INFORMATION:-}
-      - LOG_ALL_MODEL_INTERACTIONS=${LOG_ALL_MODEL_INTERACTIONS:-}
     volumes:
       - local_dynamic_storage:/home/storage
       - file_connector_tmp_storage:/home/file_connector_storage

deployment/kubernetes/api_server-service-deployment.yaml

@@ -39,23 +39,11 @@ spec:
             uvicorn danswer.main:app --host 0.0.0.0 --port 8080
         ports:
         - containerPort: 8080
-        env:
+        # There are some extra values since this is shared between services
-        - name: AUTH_TYPE
+        # There are no conflicts though, extra env variables are simply ignored
-          value: google_oauth
+        envFrom:
-        - name: POSTGRES_HOST
+        - configMapRef:
-          value: relational-db-service
+            name: env-configmap
-        - name: VESPA_HOST
-          value: document-index-service
-        - name: GOOGLE_OAUTH_CLIENT_ID
-          valueFrom:
-            secretKeyRef:
-              name: danswer-secrets
-              key: google_oauth_client_id
-        - name: GOOGLE_OAUTH_CLIENT_SECRET
-          valueFrom:
-            secretKeyRef:
-              name: danswer-secrets
-              key: google_oauth_client_secret
         volumeMounts:
         - name: dynamic-storage
           mountPath: /home/storage

deployment/kubernetes/background-deployment.yaml

@@ -17,11 +17,11 @@ spec:
         image: danswer/danswer-backend:latest
         imagePullPolicy: IfNotPresent
         command: ["/usr/bin/supervisord"]
-        env:
+        # There are some extra values since this is shared between services
-        - name: POSTGRES_HOST
+        # There are no conflicts though, extra env variables are simply ignored
-          value: relational-db-service
+        envFrom:
-        - name: VESPA_HOST
+        - configMapRef:
-          value: document-index-service
+            name: env-configmap
         volumeMounts:
         - name: dynamic-storage
           mountPath: /home/storage

deployment/kubernetes/env-configmap.yaml

@@ -0,0 +1,69 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: env-configmap
+data:
+  # Auth Setting, also check the secrets file
+  AUTH_TYPE: "disabled"  # Change this for production uses unless Danswer is only accessible behind VPN
+  SESSION_EXPIRE_TIME_SECONDS: "86400"  # 1 Day Default
+  VALID_EMAIL_DOMAINS: ""  # Can be something like danswer.ai, as an extra double-check
+  # Gen AI Settings
+  GEN_AI_MODEL_PROVIDER: "openai"
+  GEN_AI_MODEL_VERSION: "gpt-3.5-turbo"  # Use GPT-4 if you have it
+  FAST_GEN_AI_MODEL_VERSION: "gpt-3.5-turbo"
+  GEN_AI_API_KEY: ""
+  GEN_AI_API_ENDPOINT: ""
+  GEN_AI_API_VERSION: ""
+  GEN_AI_LLM_PROVIDER_TYPE: ""
+  QA_TIMEOUT: "60"
+  NUM_DOCUMENT_TOKENS_FED_TO_GENERATIVE_MODEL: ""
+  DISABLE_LLM_FILTER_EXTRACTION: ""
+  DISABLE_LLM_CHUNK_FILTER: ""
+  DISABLE_LLM_CHOOSE_SEARCH: ""
+  # Query Options
+  DOC_TIME_DECAY: ""
+  HYBRID_ALPHA: ""
+  EDIT_KEYWORD_QUERY: ""
+  MULTILINGUAL_QUERY_EXPANSION: ""
+  QA_PROMPT_OVERRIDE: ""
+  # Other Services
+  POSTGRES_HOST: "relational-db-service"
+  VESPA_HOST: "document-index-service"
+  # Don't change the NLP models unless you know what you're doing
+  DOCUMENT_ENCODER_MODEL: ""
+  NORMALIZE_EMBEDDINGS: ""
+  ASYM_QUERY_PREFIX: ""
+  ASYM_PASSAGE_PREFIX: ""
+  ENABLE_RERANKING_REAL_TIME_FLOW: ""
+  ENABLE_RERANKING_ASYNC_FLOW: ""
+  MODEL_SERVER_HOST: ""
+  MODEL_SERVER_PORT: ""
+  INDEXING_MODEL_SERVER_HOST: ""
+  MIN_THREADS_ML_MODELS: ""
+  # Indexing Configs
+  NUM_INDEXING_WORKERS: ""
+  DASK_JOB_CLIENT_ENABLED: ""
+  CONTINUE_ON_CONNECTOR_FAILURE: ""
+  EXPERIMENTAL_CHECKPOINTING_ENABLED: ""
+  CONFLUENCE_CONNECTOR_LABELS_TO_SKIP: ""
+  GONG_CONNECTOR_START_TIME: ""
+  NOTION_CONNECTOR_ENABLE_RECURSIVE_PAGE_LOOKUP: ""
+  # DanswerBot SlackBot Configs
+  DANSWER_BOT_SLACK_APP_TOKEN: ""
+  DANSWER_BOT_SLACK_BOT_TOKEN: ""
+  DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER: ""
+  DANSWER_BOT_DISPLAY_ERROR_MSGS: ""
+  DANSWER_BOT_RESPOND_EVERY_CHANNEL: ""
+  DANSWER_BOT_DISABLE_COT: ""  # Currently unused
+  NOTIFY_SLACKBOT_NO_ANSWER: ""
+  # Logging
+  # Optional Telemetry, please keep it on (nothing sensitive is collected)? <3
+  # https://docs.danswer.dev/more/telemetry
+  DISABLE_TELEMETRY: ""
+  LOG_LEVEL: ""
+  LOG_ALL_MODEL_INTERACTIONS: ""
+  LOG_VESPA_TIMING_INFORMATION: ""
+  # Shared or Non-backend Related
+  INTERNAL_URL: "http://api-server-service:80"  # for web server
+  WEB_DOMAIN: "http://localhost:3000"  # for web server and api server
+  DOMAIN: "localhost"  # for nginx

deployment/kubernetes/nginx-service-deployment.yaml

@@ -37,7 +37,10 @@ spec:
         - containerPort: 80
         env:
         - name: DOMAIN
-          value: localhost
+          valueFrom:
+            configMapKeyRef:
+              name: env-configmap
+              key: DOMAIN
         volumeMounts:
         - name: nginx-conf
           mountPath: /etc/nginx/conf.d

deployment/kubernetes/web_server-service-deployment.yaml

@@ -31,6 +31,8 @@ spec:
         imagePullPolicy: IfNotPresent
         ports:
         - containerPort: 3000
-        env:
+        # There are some extra values since this is shared between services
-        - name: INTERNAL_URL
+        # There are no conflicts though, extra env variables are simply ignored
-          value: "http://api-server-service:80"
+        envFrom:
+        - configMapRef:
+            name: env-configmap