LCORE-1422: Add streaming responses test, prow config, and cleanup

- Add streaming Responses API inline RAG e2e scenario (stream: true)
- Copy inline RAG config to e2e-prow/rhoai/configs for prow compatibility
- Extract restart into explicit "The service is restarted" Gherkin step
- Replace custom _resolve_env_var with llama_stack replace_env_vars

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: are-ces

src/llama_stack_configuration.py

@@ -6,7 +6,6 @@
 """
 
 import os
-import re
 from argparse import ArgumentParser
 from pathlib import Path
 from typing import Any, Optional
@@ -156,24 +155,6 @@ def construct_storage_backends_section(
     return output
 
 
-def _resolve_env_var(value: str) -> str:
-    """Resolve ``${env.VAR}`` and ``${env.VAR:=default}`` patterns.
-
-    Parameters:
-        value: A string that may contain an environment-variable reference.
-
-    Returns:
-        The resolved value if a matching env var is set, the default if one is
-        provided, or the original string unchanged.
-    """
-    match = re.match(r"^\$\{env\.(\w+)(?::=([^}]*))?\}$", value)
-    if match:
-        var_name = match.group(1)
-        default = match.group(2)
-        return os.environ.get(var_name, default if default is not None else value)
-    return value
-
-
 def construct_vector_stores_section(
     ls_config: dict[str, Any], byok_rag: list[dict[str, Any]]
 ) -> list[dict[str, Any]]:
@@ -206,7 +187,7 @@ def construct_vector_stores_section(
     # Resolve ${env.VAR} patterns so comparisons work when existing entries
     # use environment variable references and new entries have resolved values.
     existing_store_ids = {
-        _resolve_env_var(vs.get("vector_store_id", "")) for vs in output
+        replace_env_vars(vs.get("vector_store_id", "")) for vs in output
     }
     added = 0
     for brag in byok_rag:

tests/e2e-prow/rhoai/configs/lightspeed-stack-inline-rag.yaml

@@ -0,0 +1,38 @@
+name: Lightspeed Core Service (LCS)
+service:
+  host: 0.0.0.0
+  port: 8080
+  auth_enabled: false
+  workers: 1
+  color_log: true
+  access_log: true
+llama_stack:
+  use_as_library_client: false
+  url: http://${env.E2E_LLAMA_HOSTNAME}:8321
+  api_key: xyzzy
+user_data_collection:
+  feedback_enabled: true
+  feedback_storage: "/tmp/data/feedback"
+  transcripts_enabled: true
+  transcripts_storage: "/tmp/data/transcripts"
+
+conversation_cache:
+  type: "sqlite"
+  sqlite:
+    db_path: "/tmp/data/conversation-cache.db"
+
+authentication:
+  module: "noop"
+
+byok_rag:
+  - rag_id: e2e-test-docs
+    rag_type: inline::faiss
+    embedding_model: sentence-transformers/all-mpnet-base-v2
+    embedding_dimension: 768
+    vector_db_id: ${env.FAISS_VECTOR_STORE_ID}
+    db_path: ${env.KV_RAG_PATH:=~/.llama/storage/rag/kv_store.db}
+    score_multiplier: 1.0
+
+rag:
+  inline:
+    - e2e-test-docs

tests/e2e/features/inline_rag.feature

@@ -2,12 +2,13 @@ Feature: Inline RAG (BYOK) support tests
 
   Background:
     Given The service is started locally
+      And The system is in default state
+      And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
       And REST API service prefix is /v1
       And The service uses the lightspeed-stack-inline-rag.yaml configuration
+      And The service is restarted
 
   Scenario: Check if inline RAG source is registered
-    Given The system is in default state
-     And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
     When I access REST API endpoint rags using HTTP GET method
     Then The status code of the response is 200
      And the body of the response has the following structure
@@ -20,8 +21,6 @@ Feature: Inline RAG (BYOK) support tests
     """
 
   Scenario: Query with inline RAG returns relevant content
-    Given The system is in default state
-      And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
     When I use "query" to ask question with authorization header
     """
     {"query": "What is the title of the article from Paul?", "system_prompt": "You are an assistant. Write only lowercase letters"}
@@ -33,8 +32,6 @@ Feature: Inline RAG (BYOK) support tests
      And The response should contain non-empty rag_chunks
 
   Scenario: Inline RAG query includes referenced documents
-    Given The system is in default state
-      And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
     When I use "query" to ask question with authorization header
     """
     {"query": "What does Paul Graham say about great work?"}
@@ -43,8 +40,6 @@ Feature: Inline RAG (BYOK) support tests
      And The response should contain non-empty referenced_documents
 
   Scenario: Streaming query with inline RAG returns relevant content
-    Given The system is in default state
-      And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
     When I use "streaming_query" to ask question with authorization header
     """
     {"query": "What is the title of the article from Paul?", "system_prompt": "You are an assistant. Write only lowercase letters"}
@@ -56,8 +51,6 @@ Feature: Inline RAG (BYOK) support tests
          | great work                |
 
   Scenario: Responses API with inline RAG returns relevant content
-    Given The system is in default state
-      And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
     When I use "responses" to ask question with authorization header
     """
     {"input": "What is the title of the article from Paul?", "model": "{PROVIDER}/{MODEL}", "stream": false, "instructions": "You are an assistant. Write only lowercase letters"}
@@ -66,3 +59,14 @@ Feature: Inline RAG (BYOK) support tests
      And The response should contain following fragments
          | Fragments in LLM response |
          | great work                |
+
+  Scenario: Streaming Responses API with inline RAG returns relevant content
+    When I use "responses" to ask question with authorization header
+    """
+    {"input": "What is the title of the article from Paul?", "model": "{PROVIDER}/{MODEL}", "stream": true, "instructions": "You are an assistant. Write only lowercase letters"}
+    """
+    Then The status code of the response is 200
+     And I wait for the response to be completed
+     And The streamed response should contain following fragments
+         | Fragments in LLM response |
+         | great work                |

tests/e2e/features/steps/common.py

@@ -54,6 +54,15 @@ def configure_service(context: Context, config_name: str) -> None:
         config_path = f"tests/e2e/configuration/{mode_dir}/{config_name}"
     create_config_backup("lightspeed-stack.yaml")
     switch_config(config_path)
+
+
+@given("The service is restarted")
+def restart_service(context: Context) -> None:
+    """Restart the lightspeed-stack container and wait for it to be healthy.
+
+    Parameters:
+        context (Context): Behave context.
+    """
     restart_container("lightspeed-stack")
     # Library mode needs extra time to load embedding models after restart
     wait_for_container_health("lightspeed-stack", max_attempts=12)