From d08c2834292d3bc37691d564c15fd268a5f4541d Mon Sep 17 00:00:00 2001
From: Radovan Fuchs <rfuchs@rfuchs-thinkpadp1gen7.tpb.csb>
Date: Tue, 14 Apr 2026 08:42:23 +0200
Subject: [PATCH 1/3] improve timing

---
 tests/e2e/features/environment.py | 23 +++++++++++++++++++++++
 tests/e2e/features/steps/proxy.py | 10 +++++-----
 tests/e2e/features/steps/tls.py   |  5 ++---
 3 files changed, 30 insertions(+), 8 deletions(-)

diff --git a/tests/e2e/features/environment.py b/tests/e2e/features/environment.py
index c6e75567f..cea172333 100644
--- a/tests/e2e/features/environment.py
+++ b/tests/e2e/features/environment.py
@@ -34,6 +34,9 @@
 FALLBACK_MODEL = "gpt-4o-mini"
 FALLBACK_PROVIDER = "openai"
 
+# Wall-clock start for each feature (on ``Feature``; survives Behave context resets).
+_E2E_FEATURE_PERF_START_ATTR = "_lightspeed_e2e_feature_perf_start"
+
 
 def _fetch_models_from_service() -> dict:
     """Query /v1/models endpoint and return first LLM model.
@@ -349,7 +352,11 @@ def before_feature(context: Context, feature: Feature) -> None:
 
     Per-feature setup that is not expressed in Gherkin (e.g. feedback cleanup state).
     Lightspeed YAML is applied in feature Backgrounds via ``configure_service``.
+
+    Records monotonic start time on ``feature`` for duration logging in
+    ``after_feature`` (includes scenarios and feature teardown).
     """
+    setattr(feature, _E2E_FEATURE_PERF_START_ATTR, time.perf_counter())
     reset_active_lightspeed_stack_config_basename()
     context.active_lightspeed_stack_config_basename = None
     # One real Llama disruption per feature (module-level flag; survives context resets)
@@ -398,3 +405,19 @@ def after_feature(context: Context, feature: Feature) -> None:
 
         _stop_proxy(context, "tunnel_proxy", "proxy_loop")
         _stop_proxy(context, "interception_proxy", "interception_proxy_loop")
+
+    start = getattr(feature, _E2E_FEATURE_PERF_START_ATTR, None)
+    if start is not None:
+        elapsed_s = time.perf_counter() - start
+        try:
+            delattr(feature, _E2E_FEATURE_PERF_START_ATTR)
+        except AttributeError:
+            pass
+        feat_path = getattr(feature, "filename", "") or ""
+        label = os.path.basename(feat_path) if feat_path else feature.name
+        print(f"[e2e feature timing] {elapsed_s:.2f}s  {label}", flush=True)
+
+
+# Behave captures hook stdout by default; output is only shown in some failure paths.
+# Disable capture so feature timing lines always appear on the real console/CI log.
+after_feature.capture = False  # type: ignore[attr-defined]
diff --git a/tests/e2e/features/steps/proxy.py b/tests/e2e/features/steps/proxy.py
index 83497da23..5e1d93f9d 100644
--- a/tests/e2e/features/steps/proxy.py
+++ b/tests/e2e/features/steps/proxy.py
@@ -5,8 +5,11 @@
 The proxy sits between Llama Stack and the LLM provider (e.g., OpenAI).
 
 Config switching uses the same pattern as other e2e tests: overwrite the
-host-mounted run.yaml and restart Docker containers. Cleanup is handled
-by a Background step that restores the backup before each scenario.
+host-mounted run.yaml and restart Docker containers. Restarts are not
+triggered from ``The original Llama Stack config is restored if modified``;
+list ``Llama Stack is restarted`` / ``Lightspeed Stack is restarted`` in the
+feature file so readers see every restart. Cleanup restores the backup file
+(and stops proxy servers) before each scenario.
 """
 
 import asyncio
@@ -192,9 +195,6 @@ def restore_if_modified(context: Context) -> None:
             f"Restoring original Llama Stack config from {_LLAMA_STACK_CONFIG_BACKUP}..."
         )
         shutil.move(_LLAMA_STACK_CONFIG_BACKUP, _LLAMA_STACK_CONFIG)
-        restart_container("llama-stack")
-        restart_container("lightspeed-stack")
-        wait_for_lightspeed_stack_http_ready()
 
 
 # --- Service Restart Steps ---
diff --git a/tests/e2e/features/steps/tls.py b/tests/e2e/features/steps/tls.py
index 063405c46..c2433c8ec 100644
--- a/tests/e2e/features/steps/tls.py
+++ b/tests/e2e/features/steps/tls.py
@@ -91,9 +91,8 @@ def _configure_tls(tls_config: dict[str, Any], base_url: Optional[str] = None) -
 
 
 # --- Background Steps ---
-# Restart steps ("The original Llama Stack config is restored if modified",
-# "Llama Stack is restarted", "Lightspeed Stack is restarted") are defined in
-# proxy.py and shared across features by behave.
+# ``The original Llama Stack config is restored if modified`` only restores
+# run.yaml (see proxy.py). Restart steps are listed in tls.feature / proxy.feature.
 
 
 # --- TLS Configuration Steps ---

From 41cab3d6e38b166e2c3c351b9a07b02f50b6509f Mon Sep 17 00:00:00 2001
From: Radovan Fuchs <rfuchs@rfuchs-thinkpadp1gen7.tpb.csb>
Date: Tue, 14 Apr 2026 09:52:35 +0200
Subject: [PATCH 2/3] improve timing

---
 .github/workflows/e2e_tests.yaml                     |  2 +-
 Makefile                                             | 12 +++++++++++-
 tests/e2e/features/authorized_noop.feature           |  1 +
 tests/e2e/features/authorized_noop_token.feature     |  2 +-
 tests/e2e/features/authorized_rh_identity.feature    |  2 +-
 tests/e2e/features/conversation_cache_v2.feature     |  2 +-
 tests/e2e/features/conversations.feature             |  2 +-
 tests/e2e/features/faiss.feature                     |  2 +-
 tests/e2e/features/feedback.feature                  |  2 +-
 tests/e2e/features/health.feature                    |  1 +
 tests/e2e/features/http_401_unauthorized.feature     |  2 +-
 tests/e2e/features/info.feature                      |  1 +
 tests/e2e/features/inline_rag.feature                |  1 +
 tests/e2e/features/llama_stack_disrupted.feature     |  2 +-
 tests/e2e/features/mcp.feature                       |  1 +
 tests/e2e/features/mcp_servers_api.feature           |  2 +-
 tests/e2e/features/mcp_servers_api_auth.feature      |  2 +-
 tests/e2e/features/mcp_servers_api_no_config.feature |  2 +-
 tests/e2e/features/models.feature                    |  1 +
 tests/e2e/features/proxy.feature                     |  2 +-
 tests/e2e/features/query.feature                     |  2 +-
 tests/e2e/features/rbac.feature                      |  2 +-
 tests/e2e/features/responses.feature                 |  2 +-
 tests/e2e/features/responses_streaming.feature       |  2 +-
 tests/e2e/features/rest_api.feature                  |  1 +
 tests/e2e/features/rlsapi_v1.feature                 |  2 +-
 tests/e2e/features/rlsapi_v1_errors.feature          |  2 +-
 tests/e2e/features/smoketests.feature                |  1 +
 tests/e2e/features/streaming_query.feature           |  2 +-
 tests/e2e/features/tls.feature                       |  2 +-
 30 files changed, 40 insertions(+), 22 deletions(-)

diff --git a/.github/workflows/e2e_tests.yaml b/.github/workflows/e2e_tests.yaml
index 31d51428b..4f3960f88 100644
--- a/.github/workflows/e2e_tests.yaml
+++ b/.github/workflows/e2e_tests.yaml
@@ -11,7 +11,7 @@ jobs:
       matrix:
         mode: ["server", "library"]
         environment: ["ci"]
-    
+
     name: "E2E: ${{ matrix.mode }} mode / ${{ matrix.environment }}"
     
     env:
diff --git a/Makefile b/Makefile
index c01f93489..4ccec5212 100644
--- a/Makefile
+++ b/Makefile
@@ -32,9 +32,19 @@ test-integration: ## Run integration tests tests
 test-e2e: ## Run end to end tests for the service
 	script -q -e -c "uv run behave --color --format pretty --tags=-skip -D dump_errors=true @tests/e2e/test_list.txt"
 
-test-e2e-local: ## Run end to end tests for the service
+test-e2e-local: ## Run end to end tests for the service (no script wrapper)
 	uv run behave --color --format pretty --tags=-skip -D dump_errors=true @tests/e2e/test_list.txt
 
+# Tag-based subsets (@e2e_group_* on feature files). Default runs all groups; override for one shard, e.g.
+#   E2E_BEHAVE_TAG_EXPR='not @skip and @e2e_group_2' make test-e2e-tagged-local
+E2E_BEHAVE_TAG_EXPR ?= not @skip and (e2e_group_1 or e2e_group_2 or e2e_group_3)
+
+test-e2e-tagged: ## Run e2e tests with E2E_BEHAVE_TAG_EXPR (default: all @e2e_group_*)
+	script -q -e -c "uv run behave --color --format pretty --tags=\"$(E2E_BEHAVE_TAG_EXPR)\" -D dump_errors=true @tests/e2e/test_list.txt"
+
+test-e2e-tagged-local: ## Same as test-e2e-tagged without script wrapper
+	uv run behave --color --format pretty --tags="$(E2E_BEHAVE_TAG_EXPR)" -D dump_errors=true @tests/e2e/test_list.txt
+
 benchmarks: ## Run benchmarks
 	uv run python -m pytest -vv tests/benchmarks/
 
diff --git a/tests/e2e/features/authorized_noop.feature b/tests/e2e/features/authorized_noop.feature
index b2cdb0445..e24934f71 100644
--- a/tests/e2e/features/authorized_noop.feature
+++ b/tests/e2e/features/authorized_noop.feature
@@ -1,3 +1,4 @@
+@e2e_group_1
 Feature: Authorized endpoint API tests for the noop authentication module
 
   Background:
diff --git a/tests/e2e/features/authorized_noop_token.feature b/tests/e2e/features/authorized_noop_token.feature
index a1d4c80f6..e8f75d2f8 100644
--- a/tests/e2e/features/authorized_noop_token.feature
+++ b/tests/e2e/features/authorized_noop_token.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_2 @Authorized
 Feature: Authorized endpoint API tests for the noop-with-token authentication module
 
   Background:
diff --git a/tests/e2e/features/authorized_rh_identity.feature b/tests/e2e/features/authorized_rh_identity.feature
index a4ae18289..3e196527d 100644
--- a/tests/e2e/features/authorized_rh_identity.feature
+++ b/tests/e2e/features/authorized_rh_identity.feature
@@ -1,4 +1,4 @@
-@RHIdentity
+@e2e_group_3 @RHIdentity
 Feature: Authorized endpoint API tests for the rh-identity authentication module
 
   Background:
diff --git a/tests/e2e/features/conversation_cache_v2.feature b/tests/e2e/features/conversation_cache_v2.feature
index 291d7ed96..7b10e9114 100644
--- a/tests/e2e/features/conversation_cache_v2.feature
+++ b/tests/e2e/features/conversation_cache_v2.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_2 @Authorized
 Feature: Conversation Cache V2 API tests
 
   Background:
diff --git a/tests/e2e/features/conversations.feature b/tests/e2e/features/conversations.feature
index 27c7c94b7..8f47cfb24 100644
--- a/tests/e2e/features/conversations.feature
+++ b/tests/e2e/features/conversations.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_2 @Authorized
 Feature: conversations endpoint API tests
 
   Background:
diff --git a/tests/e2e/features/faiss.feature b/tests/e2e/features/faiss.feature
index d321cbe53..db50e4d28 100644
--- a/tests/e2e/features/faiss.feature
+++ b/tests/e2e/features/faiss.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_1 @Authorized
 Feature: FAISS support tests
 
   Background:
diff --git a/tests/e2e/features/feedback.feature b/tests/e2e/features/feedback.feature
index 11ed24f8e..6edc1f079 100644
--- a/tests/e2e/features/feedback.feature
+++ b/tests/e2e/features/feedback.feature
@@ -1,4 +1,4 @@
-@Feedback
+@e2e_group_3 @Feedback
 Feature: feedback endpoint API tests
 
 
diff --git a/tests/e2e/features/health.feature b/tests/e2e/features/health.feature
index 53f222e2e..17846d07c 100644
--- a/tests/e2e/features/health.feature
+++ b/tests/e2e/features/health.feature
@@ -1,3 +1,4 @@
+@e2e_group_2
 Feature: REST API tests
 
 
diff --git a/tests/e2e/features/http_401_unauthorized.feature b/tests/e2e/features/http_401_unauthorized.feature
index aafd19967..435e190bb 100644
--- a/tests/e2e/features/http_401_unauthorized.feature
+++ b/tests/e2e/features/http_401_unauthorized.feature
@@ -1,4 +1,4 @@
-@Authorized @Feedback @RHIdentity @RBAC
+@e2e_group_3 @Authorized @Feedback @RHIdentity @RBAC
 Feature: HTTP 401 Unauthorized
 
   Aggregates end-to-end scenarios that assert a 401 response when authentication
diff --git a/tests/e2e/features/info.feature b/tests/e2e/features/info.feature
index 3fa561ff7..b3b0472bb 100644
--- a/tests/e2e/features/info.feature
+++ b/tests/e2e/features/info.feature
@@ -1,3 +1,4 @@
+@e2e_group_3
 Feature: Info tests
 
 
diff --git a/tests/e2e/features/inline_rag.feature b/tests/e2e/features/inline_rag.feature
index 194af71a6..53da476ab 100644
--- a/tests/e2e/features/inline_rag.feature
+++ b/tests/e2e/features/inline_rag.feature
@@ -1,3 +1,4 @@
+@e2e_group_3
 Feature: Inline RAG (BYOK) support tests
 
   Background:
diff --git a/tests/e2e/features/llama_stack_disrupted.feature b/tests/e2e/features/llama_stack_disrupted.feature
index cb1afccda..d7e28877c 100644
--- a/tests/e2e/features/llama_stack_disrupted.feature
+++ b/tests/e2e/features/llama_stack_disrupted.feature
@@ -1,4 +1,4 @@
-@skip-in-library-mode @Authorized
+@e2e_group_3 @skip-in-library-mode @Authorized
 Feature: Llama Stack connection disrupted
 
   End-to-end scenarios that stop the Llama Stack container (or simulate disconnect) and
diff --git a/tests/e2e/features/mcp.feature b/tests/e2e/features/mcp.feature
index e5b1b6b7b..ee704ca37 100644
--- a/tests/e2e/features/mcp.feature
+++ b/tests/e2e/features/mcp.feature
@@ -1,3 +1,4 @@
+@e2e_group_2
 Feature: MCP tests
 
   Background:
diff --git a/tests/e2e/features/mcp_servers_api.feature b/tests/e2e/features/mcp_servers_api.feature
index 73c4ed381..7d7de8c28 100644
--- a/tests/e2e/features/mcp_servers_api.feature
+++ b/tests/e2e/features/mcp_servers_api.feature
@@ -1,4 +1,4 @@
-@MCP
+@e2e_group_3 @MCP
 Feature: MCP Server Management API tests
 
   Tests for the dynamic MCP server management endpoints:
diff --git a/tests/e2e/features/mcp_servers_api_auth.feature b/tests/e2e/features/mcp_servers_api_auth.feature
index 106e93bcd..e28e475fd 100644
--- a/tests/e2e/features/mcp_servers_api_auth.feature
+++ b/tests/e2e/features/mcp_servers_api_auth.feature
@@ -1,4 +1,4 @@
-@MCPServerAPIAuth
+@e2e_group_1 @MCPServerAPIAuth
 Feature: MCP Server Management API authentication tests
 
   Tests that the MCP server management endpoints enforce authentication
diff --git a/tests/e2e/features/mcp_servers_api_no_config.feature b/tests/e2e/features/mcp_servers_api_no_config.feature
index 663592601..896873350 100644
--- a/tests/e2e/features/mcp_servers_api_no_config.feature
+++ b/tests/e2e/features/mcp_servers_api_no_config.feature
@@ -1,4 +1,4 @@
-@MCPNoConfig
+@e2e_group_1 @MCPNoConfig
 Feature: MCP Server API tests without configured MCP servers
 
   Tests that the MCP server management endpoints work correctly
diff --git a/tests/e2e/features/models.feature b/tests/e2e/features/models.feature
index da7b98613..92c9fca2b 100644
--- a/tests/e2e/features/models.feature
+++ b/tests/e2e/features/models.feature
@@ -1,3 +1,4 @@
+@e2e_group_2
 Feature: Models endpoint tests
 
 
diff --git a/tests/e2e/features/proxy.feature b/tests/e2e/features/proxy.feature
index 19daba1cd..aaab54f4e 100644
--- a/tests/e2e/features/proxy.feature
+++ b/tests/e2e/features/proxy.feature
@@ -1,4 +1,4 @@
-@skip-in-library-mode
+@e2e_group_3 @skip-in-library-mode
 Feature: Proxy and TLS networking tests for Llama Stack providers
 
   Verify that the Lightspeed Stack works correctly when Llama Stack's
diff --git a/tests/e2e/features/query.feature b/tests/e2e/features/query.feature
index e1b7e4112..b91dae7fe 100644
--- a/tests/e2e/features/query.feature
+++ b/tests/e2e/features/query.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_3 @Authorized
 Feature: Query endpoint API tests
 
   Background:
diff --git a/tests/e2e/features/rbac.feature b/tests/e2e/features/rbac.feature
index bf67e7cc9..07d711ddb 100644
--- a/tests/e2e/features/rbac.feature
+++ b/tests/e2e/features/rbac.feature
@@ -1,4 +1,4 @@
-@RBAC
+@e2e_group_2 @RBAC
 Feature: Role-Based Access Control (RBAC)
 
   Comprehensive tests for role-based access control to ensure
diff --git a/tests/e2e/features/responses.feature b/tests/e2e/features/responses.feature
index 8568b5c3f..cfb9fbecc 100644
--- a/tests/e2e/features/responses.feature
+++ b/tests/e2e/features/responses.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_1 @Authorized
 Feature: Responses endpoint API tests
 
   Background:
diff --git a/tests/e2e/features/responses_streaming.feature b/tests/e2e/features/responses_streaming.feature
index 17e2edd14..2e511f4a0 100644
--- a/tests/e2e/features/responses_streaming.feature
+++ b/tests/e2e/features/responses_streaming.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_1 @Authorized
 Feature: Responses endpoint streaming API tests
 
 # Same coverage as ``responses.feature`` with ``stream=true`` (SSE for success paths;
diff --git a/tests/e2e/features/rest_api.feature b/tests/e2e/features/rest_api.feature
index f1a14544e..a40bfd248 100644
--- a/tests/e2e/features/rest_api.feature
+++ b/tests/e2e/features/rest_api.feature
@@ -1,3 +1,4 @@
+@e2e_group_1
 Feature: REST API tests
 
 
diff --git a/tests/e2e/features/rlsapi_v1.feature b/tests/e2e/features/rlsapi_v1.feature
index d0567d24c..5cd7ce34c 100644
--- a/tests/e2e/features/rlsapi_v1.feature
+++ b/tests/e2e/features/rlsapi_v1.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_2 @Authorized
 Feature: rlsapi v1 /infer endpoint API tests
 
   Background:
diff --git a/tests/e2e/features/rlsapi_v1_errors.feature b/tests/e2e/features/rlsapi_v1_errors.feature
index b4618fed5..8dbe3c9c7 100644
--- a/tests/e2e/features/rlsapi_v1_errors.feature
+++ b/tests/e2e/features/rlsapi_v1_errors.feature
@@ -1,4 +1,4 @@
-@RBAC
+@e2e_group_1 @RBAC
 Feature: rlsapi v1 /infer endpoint error response tests
 
   Tests for error conditions on the rlsapi v1 /infer endpoint including
diff --git a/tests/e2e/features/smoketests.feature b/tests/e2e/features/smoketests.feature
index 1e57df982..039d54d0d 100644
--- a/tests/e2e/features/smoketests.feature
+++ b/tests/e2e/features/smoketests.feature
@@ -1,3 +1,4 @@
+@e2e_group_3
 Feature: Smoke tests
 
 
diff --git a/tests/e2e/features/streaming_query.feature b/tests/e2e/features/streaming_query.feature
index 3576145e7..c7567442f 100644
--- a/tests/e2e/features/streaming_query.feature
+++ b/tests/e2e/features/streaming_query.feature
@@ -1,4 +1,4 @@
-@Authorized
+@e2e_group_2 @Authorized
 Feature: streaming_query endpoint API tests
 
   Background:
diff --git a/tests/e2e/features/tls.feature b/tests/e2e/features/tls.feature
index a32763bd1..5a2d77338 100644
--- a/tests/e2e/features/tls.feature
+++ b/tests/e2e/features/tls.feature
@@ -1,4 +1,4 @@
-@skip-in-library-mode
+@e2e_group_1 @skip-in-library-mode
 Feature: TLS configuration for remote inference providers
   Validate that Llama Stack's NetworkConfig.tls settings are applied correctly
   when connecting to a remote inference provider over HTTPS.

From 3fd37d759bcbcf67c5ee30c136b2f2925e389dbf Mon Sep 17 00:00:00 2001
From: Radovan Fuchs <rfuchs@rfuchs-thinkpadp1gen7.tpb.csb>
Date: Tue, 14 Apr 2026 09:57:59 +0200
Subject: [PATCH 3/3] improve timing

---
 .github/workflows/e2e_tests.yaml | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/e2e_tests.yaml b/.github/workflows/e2e_tests.yaml
index 4f3960f88..c2413740a 100644
--- a/.github/workflows/e2e_tests.yaml
+++ b/.github/workflows/e2e_tests.yaml
@@ -11,8 +11,9 @@ jobs:
       matrix:
         mode: ["server", "library"]
         environment: ["ci"]
+        e2e_group: [1, 2, 3]
 
-    name: "E2E: ${{ matrix.mode }} mode / ${{ matrix.environment }}"
+    name: "E2E: ${{ matrix.mode }} mode / ${{ matrix.environment }} / group ${{ matrix.e2e_group }}"
     
     env:
       OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
@@ -103,6 +104,7 @@ jobs:
           echo "=== Configuration Summary ==="
           echo "Deployment mode: ${{ matrix.mode }}"
           echo "Environment: ${{ matrix.environment }}"
+          echo "E2E shard (Makefile test-e2e-tagged): @e2e_group_${{ matrix.e2e_group }} (with not @skip)"
           echo "Source config: tests/e2e/configs/run-${{ matrix.environment }}.yaml"
           echo ""
           echo "=== Configuration Preview ==="
@@ -187,13 +189,15 @@ jobs:
           TERM: xterm-256color
           FORCE_COLOR: 1
           E2E_DEPLOYMENT_MODE: ${{ matrix.mode }}
+          # Matches Makefile test-e2e-tagged / E2E_BEHAVE_TAG_EXPR (one @e2e_group_* per job).
+          E2E_BEHAVE_TAG_EXPR: "not @skip and @e2e_group_${{ matrix.e2e_group }}"
         run: |
           echo "Installing test dependencies..."
           pip install uv
           uv sync
 
-          echo "Running comprehensive e2e test suite..."
-          make test-e2e
+          echo "Running e2e tests (E2E_BEHAVE_TAG_EXPR=${E2E_BEHAVE_TAG_EXPR})..."
+          make test-e2e-tagged
 
       - name: Show logs on failure
         if: failure()