showcase using env vars

distributions/remote-vllm/compose.yaml

@@ -71,6 +71,13 @@ services:
       - ~/.llama:/root/.llama
       - ~/local/llama-stack/distributions/remote-vllm/run.yaml:/root/llamastack-run-remote-vllm.yaml
     # network_mode: "host"
+    environment:
+      - LLAMA_INFERENCE_VLLM_URL=${LLAMA_INFERENCE_VLLM_URL:-http://host.docker.internal:5100/v1}
+      - LLAMA_INFERENCE_MODEL=${LLAMA_INFERENCE_MODEL:-Llama3.1-8B-Instruct}
+      - MAX_TOKENS=${MAX_TOKENS:-4096}
+      - SQLITE_STORE_DIR=${SQLITE_STORE_DIR:-$HOME/.llama/distributions/remote-vllm}
+      - LLAMA_SAFETY_VLLM_URL=${LLAMA_SAFETY_VLLM_URL:-http://host.docker.internal:5101/v1}
+      - LLAMA_SAFETY_MODEL=${LLAMA_SAFETY_MODEL:-Llama-Guard-3-1B}
     ports:
       - "5001:5001"
     # Hack: wait for vLLM server to start before starting docker

distributions/remote-vllm/run.yaml

@@ -16,16 +16,16 @@ providers:
     provider_type: remote::vllm
     config:
       # NOTE: replace with "localhost" if you are running in "host" network mode
-      url: http://host.docker.internal:5100/v1
+      url: ${env.LLAMA_INFERENCE_VLLM_URL:http://host.docker.internal:5100/v1}
-      max_tokens: 4096
+      max_tokens: ${env.MAX_TOKENS:4096}
       api_token: fake
   # serves safety llama_guard model
   - provider_id: vllm-1
     provider_type: remote::vllm
     config:
       # NOTE: replace with "localhost" if you are running in "host" network mode
-      url: http://host.docker.internal:5101/v1
+      url: ${env.LLAMA_SAFETY_VLLM_URL:http://host.docker.internal:5101/v1}
-      max_tokens: 4096
+      max_tokens: ${env.MAX_TOKENS:4096}
       api_token: fake
   memory:
   - provider_id: faiss-0
@@ -34,7 +34,7 @@ providers:
       kvstore:
         namespace: null
         type: sqlite
-        db_path: /home/ashwin/.llama/distributions/remote-vllm/faiss_store.db
+        db_path: "${env.SQLITE_STORE_DIR:/home/ashwin/.llama/distributions/remote-vllm}/faiss_store.db"
   safety:
   - provider_id: llama-guard
     provider_type: inline::llama-guard
@@ -50,7 +50,7 @@ providers:
       persistence_store:
         namespace: null
         type: sqlite
-        db_path: /home/ashwin/.llama/distributions/remote-vllm/agents_store.db
+        db_path: "${env.SQLITE_STORE_DIR:/home/ashwin/.llama/distributions/remote-vllm}/agents_store.db"
   telemetry:
   - provider_id: meta0
     provider_type: inline::meta-reference
@@ -58,11 +58,11 @@ providers:
 metadata_store:
   namespace: null
   type: sqlite
-  db_path: /home/ashwin/.llama/distributions/remote-vllm/registry.db
+  db_path: "${env.SQLITE_STORE_DIR:/home/ashwin/.llama/distributions/remote-vllm}/registry.db"
 models:
-  - model_id: Llama3.1-8B-Instruct
+  - model_id: ${env.LLAMA_INFERENCE_MODEL:Llama3.1-8B-Instruct}
     provider_id: vllm-0
-  - model_id: Llama-Guard-3-1B
+  - model_id: ${env.LLAMA_SAFETY_MODEL:Llama-Guard-3-1B}
     provider_id: vllm-1
 shields:
-  - shield_id: Llama-Guard-3-1B
+  - shield_id: ${env.LLAMA_SAFETY_MODEL:Llama-Guard-3-1B}

llama_stack/distribution/server/server.py

@@ -295,11 +295,17 @@ def replace_env_vars(config: Any, path: str = "") -> Any:
             env_var = match.group(1)
             default_val = match.group(2)
 
-            if env_var not in os.environ:
+            if default_val is None:
-                if default_val is None:
+                if env_var not in os.environ:
                     raise EnvVarError(env_var, path)
-                return default_val
+                value = os.environ[env_var]
-            return os.environ[env_var]
+            else:
+                # use the default if env var is "nullish"
+                value = os.environ.get(env_var)
+                if not value:
+                    value = default_val
+
+            return value
 
         try:
             return re.sub(pattern, get_env_var, config)