fix(api): enforce global ollama endpoint order

apps/api/src/services/embedding_service.py

@@ -22,7 +22,7 @@ import httpx
 import structlog
 
 from src.services.model_registry import get_model as _get_model
-from src.services.ollama_endpoint_resolver import resolve_ollama_endpoint
+from src.services.ollama_endpoint_resolver import resolve_ollama_order
 
 logger = structlog.get_logger(__name__)
 
@@ -98,7 +98,15 @@ class OllamaEmbeddingService:
         P1 修復 (2026-03-29): 維度配置化，支援更多模型
         """
         self._model = model
-        self._ollama_url = ollama_url or resolve_ollama_endpoint("embedding")
+        if ollama_url:
+            self._ollama_endpoints = ((ollama_url, "custom"),)
+        else:
+            self._ollama_endpoints = tuple(
+                (endpoint.url, endpoint.provider_name)
+                for endpoint in resolve_ollama_order("embedding")
+                if endpoint.url
+            )
+        self._ollama_url = self._ollama_endpoints[0][0] if self._ollama_endpoints else ""
         self._timeout = timeout
         self._default_dimension = default_dimension or self.MODEL_DIMENSIONS.get(
             model, self.DEFAULT_DIMENSION
@@ -142,43 +150,65 @@ class OllamaEmbeddingService:
         """
         client = await self._get_client()
 
-        try:
-            response = await client.post(
-                f"{self._ollama_url}/api/embeddings",
-                json={
-                    "model": self._model,
-                    "prompt": text,
-                },
-            )
-            response.raise_for_status()
+        last_error: Exception | None = None
+        for endpoint_url, provider_name in self._ollama_endpoints:
+            try:
+                response = await client.post(
+                    f"{endpoint_url}/api/embeddings",
+                    json={
+                        "model": self._model,
+                        "prompt": text,
+                    },
+                )
+                response.raise_for_status()
 
-            data = response.json()
-            embedding = data.get("embedding", [])
+                data = response.json()
+                embedding = data.get("embedding", [])
 
-            # 更新維度快取
-            if self._dimension is None and embedding:
-                self._dimension = len(embedding)
-                logger.info(
-                    "embedding_dimension_detected",
+                # 更新維度快取
+                if self._dimension is None and embedding:
+                    self._dimension = len(embedding)
+                    logger.info(
+                        "embedding_dimension_detected",
+                        model=self._model,
+                        dimension=self._dimension,
+                        provider=provider_name,
+                    )
+
+                return embedding
+
+            except httpx.TimeoutException as e:
+                last_error = e
+                logger.error(
+                    "embedding_timeout",
                     model=self._model,
-                    dimension=self._dimension,
+                    text_len=len(text),
+                    provider=provider_name,
+                )
+            except httpx.HTTPStatusError as e:
+                last_error = e
+                logger.error(
+                    "embedding_http_error",
+                    status=e.response.status_code,
+                    model=self._model,
+                    provider=provider_name,
+                )
+            except Exception as e:
+                last_error = e
+                logger.error(
+                    "embedding_error",
+                    error=str(e),
+                    model=self._model,
+                    provider=provider_name,
                 )
 
-            return embedding
-
-        except httpx.TimeoutException as e:
-            logger.error("embedding_timeout", model=self._model, text_len=len(text))
-            raise EmbeddingError(f"Embedding timeout after {self._timeout}s") from e
-        except httpx.HTTPStatusError as e:
-            logger.error(
-                "embedding_http_error",
-                status=e.response.status_code,
-                model=self._model,
-            )
-            raise EmbeddingError(f"Ollama API error: {e.response.status_code}") from e
-        except Exception as e:
-            logger.error("embedding_error", error=str(e), model=self._model)
-            raise EmbeddingError(f"Embedding failed: {e}") from e
+        if isinstance(last_error, httpx.TimeoutException):
+            raise EmbeddingError(f"Embedding timeout after {self._timeout}s") from last_error
+        if isinstance(last_error, httpx.HTTPStatusError):
+            raise EmbeddingError(
+                f"Ollama API error: {last_error.response.status_code}"
+            ) from last_error
+        raise EmbeddingError("Embedding failed on all Ollama endpoints") from last_error
 
     async def embed_batch(
         self,

apps/api/src/services/knowledge_extractor_service.py

@@ -4,7 +4,7 @@ Knowledge Extractor Service — KB Phase 2-A
 Incident resolved 後自動萃取 KB 草稿。
 
 設計原則:
-- 強制使用 Ollama llama3.2:3b（本地推理，符合 Phase 24 D7 隱私規則）
+- 使用 Ollama llama3.2:3b，依全域順序 GCP-A → GCP-B → 111 嘗試
 - fire-and-forget：失敗不影響 resolve 主流程
 - logger.exception 保留完整 Stack Trace 供 Prompt 調優
 
@@ -15,11 +15,11 @@ import structlog
 
 logger = structlog.get_logger(__name__)
 
-# 2026-05-05 Codex: KB 萃取走 111 lane，避免污染 GCP alert-fast lane
-def _get_ollama_base() -> str:
-    from src.services.ollama_endpoint_resolver import resolve_ollama_endpoint
+# 2026-05-19 Codex: 統帥校正，全 Ollama workload 固定 GCP-A → GCP-B → 111。
+def _get_ollama_endpoints():
+    from src.services.ollama_endpoint_resolver import resolve_ollama_order
 
-    return resolve_ollama_endpoint("deep_rca")
+    return resolve_ollama_order("deep_rca")
 _EXTRACT_MODEL = "llama3.2:3b"
 _EXTRACT_TIMEOUT = 30.0  # 秒，容忍慢速
 
@@ -160,36 +160,54 @@ class KnowledgeExtractorService:
 
         不走 AIRouter 是刻意設計：
         - KB 萃取是背景工作，不需要完整的路由/閘門/Cache 邏輯
-        - 強制本地，不允許 fallback 到 cloud provider
+        - Ollama endpoint 固定依 GCP-A → GCP-B → 111 嘗試
         """
         import httpx
 
-        try:
-            async with httpx.AsyncClient(timeout=_EXTRACT_TIMEOUT) as client:
-                r = await client.post(
-                    f"{_get_ollama_base()}/api/generate",
-                    json={
-                        "model": _EXTRACT_MODEL,
-                        "prompt": prompt,
-                        "stream": False,
-                        "options": {
-                            "temperature": 0.3,   # 低溫：減少幻覺
-                            "num_predict": 800,   # 控制長度
-                            "stop": ["\n\n\n"],   # 防止無限生成
+        endpoints = _get_ollama_endpoints()
+        async with httpx.AsyncClient(timeout=_EXTRACT_TIMEOUT) as client:
+            for endpoint in endpoints:
+                if not endpoint.url:
+                    continue
+                try:
+                    r = await client.post(
+                        f"{endpoint.url}/api/generate",
+                        json={
+                            "model": _EXTRACT_MODEL,
+                            "prompt": prompt,
+                            "stream": False,
+                            "options": {
+                                "temperature": 0.3,   # 低溫：減少幻覺
+                                "num_predict": 800,   # 控制長度
+                                "stop": ["\n\n\n"],   # 防止無限生成
+                            },
                         },
-                    },
-                )
-                r.raise_for_status()
-                text = r.json().get("response", "").strip()
-                return text or None
+                    )
+                    r.raise_for_status()
+                    text = r.json().get("response", "").strip()
+                    if text:
+                        logger.info(
+                            "kb_ollama_call_success",
+                            model=_EXTRACT_MODEL,
+                            provider=endpoint.provider_name,
+                            base=endpoint.url,
+                        )
+                        return text
+                except Exception as e:
+                    logger.warning(
+                        "kb_ollama_call_failed",
+                        model=_EXTRACT_MODEL,
+                        provider=endpoint.provider_name,
+                        base=endpoint.url,
+                        error=str(e),
+                    )
 
-        except Exception:
-            logger.exception(
-                "kb_ollama_call_failed",
-                model=_EXTRACT_MODEL,
-                base=_get_ollama_base(),
-            )
-            return None
+        logger.error(
+            "kb_ollama_all_endpoints_failed",
+            model=_EXTRACT_MODEL,
+            attempted=[endpoint.provider_name for endpoint in endpoints],
+        )
+        return None
 
     def _extract_title(self, markdown: str, incident) -> str:
         """

apps/api/src/services/knowledge_rag_service.py

@@ -8,7 +8,7 @@ AWOOOI — Knowledge RAG Service (Phase 33, ADR-067)
   - 超過 100 筆: 執行 CREATE INDEX ivfflat (手動觸發)
 
 向量模型: bge-m3 (GCP-A/GCP-B/111 Ollama lane, 1024維)
-生成模型: qwen2.5:7b-instruct (Ollama 111)
+生成模型: qwen2.5:7b-instruct (Ollama GCP-A/GCP-B/111)
 
 leWOOOgo: Service 層只處理業務邏輯，DB 存取委派 rag_chunk_repository
 架構審查 C1 修正: 2026-04-10 Claude Sonnet 4.6 Asia/Taipei
@@ -22,7 +22,7 @@ import structlog
 
 import src.repositories.rag_chunk_repository as rag_repo
 from src.core.config import settings
-from src.services.ollama_endpoint_resolver import resolve_ollama_endpoint
+from src.services.ollama_endpoint_resolver import resolve_ollama_order
 
 logger = structlog.get_logger(__name__)
 
@@ -128,19 +128,35 @@ class KnowledgeRAGService:
     # ------------------------------------------------------------------
 
     async def _embed(self, text: str) -> list[float] | None:
-        try:
-            http = await self._get_http()
-            resp = await http.post(
-                f"{resolve_ollama_endpoint('embedding')}/api/embeddings",
-                json={
-                    "model": getattr(settings, "OLLAMA_EMBEDDING_MODEL", _EMBED_MODEL),
-                    "prompt": text,
-                },
-            )
-            if resp.status_code == 200:
-                return resp.json().get("embedding")
-        except Exception as e:
-            logger.warning("rag_embed_failed", error=str(e))
+        http = await self._get_http()
+        for endpoint in resolve_ollama_order("embedding"):
+            if not endpoint.url:
+                continue
+            try:
+                resp = await http.post(
+                    f"{endpoint.url}/api/embeddings",
+                    json={
+                        "model": getattr(settings, "OLLAMA_EMBEDDING_MODEL", _EMBED_MODEL),
+                        "prompt": text,
+                    },
+                )
+                if resp.status_code == 200:
+                    logger.debug(
+                        "rag_embed_success",
+                        provider=endpoint.provider_name,
+                    )
+                    return resp.json().get("embedding")
+                logger.warning(
+                    "rag_embed_http_error",
+                    provider=endpoint.provider_name,
+                    status=resp.status_code,
+                )
+            except Exception as e:
+                logger.warning(
+                    "rag_embed_failed",
+                    provider=endpoint.provider_name,
+                    error=str(e),
+                )
         return None
 
     async def _generate_answer(self, question: str, context: str) -> str:
@@ -150,22 +166,38 @@ class KnowledgeRAGService:
             f"=== 相關資料 ===\n{context[:6000]}\n\n"
             f"=== 問題 ===\n{question}"
         )
-        try:
-            http = await self._get_http()
-            resp = await http.post(
-                f"{resolve_ollama_endpoint('rag')}/api/generate",
-                json={
-                    "model": _GEN_MODEL,
-                    "prompt": prompt,
-                    "stream": False,
-                    "options": {"num_predict": 512, "temperature": 0.2},
-                },
-                timeout=httpx.Timeout(90.0, connect=10.0),
-            )
-            if resp.status_code == 200:
-                return resp.json().get("response", "").strip()
-        except Exception as e:
-            logger.error("rag_generate_failed", error=str(e))
+        http = await self._get_http()
+        for endpoint in resolve_ollama_order("rag"):
+            if not endpoint.url:
+                continue
+            try:
+                resp = await http.post(
+                    f"{endpoint.url}/api/generate",
+                    json={
+                        "model": _GEN_MODEL,
+                        "prompt": prompt,
+                        "stream": False,
+                        "options": {"num_predict": 512, "temperature": 0.2},
+                    },
+                    timeout=httpx.Timeout(90.0, connect=10.0),
+                )
+                if resp.status_code == 200:
+                    logger.debug(
+                        "rag_generate_success",
+                        provider=endpoint.provider_name,
+                    )
+                    return resp.json().get("response", "").strip()
+                logger.warning(
+                    "rag_generate_http_error",
+                    provider=endpoint.provider_name,
+                    status=resp.status_code,
+                )
+            except Exception as e:
+                logger.error(
+                    "rag_generate_failed",
+                    provider=endpoint.provider_name,
+                    error=str(e),
+                )
         return "⚠️ RAG 生成失敗，請稍後再試"
 
     # ------------------------------------------------------------------

apps/api/src/services/local_code_review_service.py

@@ -20,7 +20,7 @@ import structlog
 
 from src.core.config import get_settings
 from src.services.model_registry import get_model
-from src.services.ollama_endpoint_resolver import resolve_ollama_endpoint
+from src.services.ollama_endpoint_resolver import resolve_ollama_order
 
 logger = structlog.get_logger(__name__)
 settings = get_settings()
@@ -120,26 +120,54 @@ class LocalCodeReviewService:
             "格式：每個問題獨立一行，以「⚠️」開頭。如果沒有問題，說「✅ 程式碼品質良好」\n\n"
             f"=== Diff ===\n{diff[:40000]}\n=== 結束 ==="
         )
-        try:
-            http = await self._get_http()
-            resp = await http.post(
-                f"{resolve_ollama_endpoint('code_review')}/api/generate",
-                json={
-                    "model": _MODEL_OLLAMA,
-                    "prompt": prompt,
-                    "stream": False,
-                    "options": {"num_predict": 1024, "temperature": 0.1},
-                },
-            )
-            if resp.status_code == 200:
-                text = resp.json().get("response", "").strip()
-                issues = text.count("⚠️")
-                logger.info("pr_review_ollama_done", pr_id=pr_id, issues=issues)
-                return {"review_text": text, "issues_count": issues, "model": _MODEL_OLLAMA, "provider": "ollama"}
-        except httpx.TimeoutException:
-            logger.warning("pr_review_ollama_timeout", pr_id=pr_id)
-        except Exception as e:
-            logger.error("pr_review_ollama_failed", pr_id=pr_id, error=str(e))
+        http = await self._get_http()
+        for endpoint in resolve_ollama_order("code_review"):
+            if not endpoint.url:
+                continue
+            try:
+                resp = await http.post(
+                    f"{endpoint.url}/api/generate",
+                    json={
+                        "model": _MODEL_OLLAMA,
+                        "prompt": prompt,
+                        "stream": False,
+                        "options": {"num_predict": 1024, "temperature": 0.1},
+                    },
+                )
+                if resp.status_code == 200:
+                    text = resp.json().get("response", "").strip()
+                    issues = text.count("⚠️")
+                    logger.info(
+                        "pr_review_ollama_done",
+                        pr_id=pr_id,
+                        issues=issues,
+                        provider=endpoint.provider_name,
+                    )
+                    return {
+                        "review_text": text,
+                        "issues_count": issues,
+                        "model": _MODEL_OLLAMA,
+                        "provider": endpoint.provider_name,
+                    }
+                logger.warning(
+                    "pr_review_ollama_http_error",
+                    pr_id=pr_id,
+                    provider=endpoint.provider_name,
+                    status=resp.status_code,
+                )
+            except httpx.TimeoutException:
+                logger.warning(
+                    "pr_review_ollama_timeout",
+                    pr_id=pr_id,
+                    provider=endpoint.provider_name,
+                )
+            except Exception as e:
+                logger.error(
+                    "pr_review_ollama_failed",
+                    pr_id=pr_id,
+                    provider=endpoint.provider_name,
+                    error=str(e),
+                )
         return None
 
     async def _review_with_gemini(
@@ -243,26 +271,55 @@ class LocalCodeReviewService:
                 "格式：每個問題以「⚠️」開頭，如無問題說「✅ Push 品質正常」\n"
                 "5 行以內，繁體中文。"
             )
-            try:
-                http = await self._get_http()
-                resp = await http.post(
-                    f"{resolve_ollama_endpoint('code_review')}/api/generate",
-                    json={
-                        "model": _MODEL_OLLAMA,
-                        "prompt": prompt,
-                        "stream": False,
-                        "options": {"num_predict": 512, "temperature": 0.1},
-                    },
-                )
-                if resp.status_code == 200:
-                    text = resp.json().get("response", "").strip()
-                    issues = text.count("⚠️")
-                    logger.info("push_review_ollama_done", repo=repo_name, branch=branch, issues=issues)
-                    return {"review_text": text, "issues_count": issues, "model": _MODEL_OLLAMA, "provider": "ollama"}
-            except httpx.TimeoutException:
-                logger.warning("push_review_ollama_timeout", repo=repo_name)
-            except Exception as e:
-                logger.error("push_review_ollama_failed", repo=repo_name, error=str(e))
+            http = await self._get_http()
+            for endpoint in resolve_ollama_order("code_review"):
+                if not endpoint.url:
+                    continue
+                try:
+                    resp = await http.post(
+                        f"{endpoint.url}/api/generate",
+                        json={
+                            "model": _MODEL_OLLAMA,
+                            "prompt": prompt,
+                            "stream": False,
+                            "options": {"num_predict": 512, "temperature": 0.1},
+                        },
+                    )
+                    if resp.status_code == 200:
+                        text = resp.json().get("response", "").strip()
+                        issues = text.count("⚠️")
+                        logger.info(
+                            "push_review_ollama_done",
+                            repo=repo_name,
+                            branch=branch,
+                            issues=issues,
+                            provider=endpoint.provider_name,
+                        )
+                        return {
+                            "review_text": text,
+                            "issues_count": issues,
+                            "model": _MODEL_OLLAMA,
+                            "provider": endpoint.provider_name,
+                        }
+                    logger.warning(
+                        "push_review_ollama_http_error",
+                        repo=repo_name,
+                        provider=endpoint.provider_name,
+                        status=resp.status_code,
+                    )
+                except httpx.TimeoutException:
+                    logger.warning(
+                        "push_review_ollama_timeout",
+                        repo=repo_name,
+                        provider=endpoint.provider_name,
+                    )
+                except Exception as e:
+                    logger.error(
+                        "push_review_ollama_failed",
+                        repo=repo_name,
+                        provider=endpoint.provider_name,
+                        error=str(e),
+                    )
             return None
 
     async def close(self) -> None:

apps/api/src/services/ollama_endpoint_resolver.py

@@ -1,9 +1,9 @@
 """
-Ollama endpoint resolver for non-critical workload placement.
+Ollama endpoint resolver for AWOOOI workload placement.
 
-ADR-110 gives AWOOOI three Ollama endpoints. This resolver is intentionally
-small: it chooses the preferred endpoint by workload class, while health-aware
-failover remains owned by ollama_failover_manager.py.
+ADR-110 gives AWOOOI three Ollama endpoints. The global order is always
+GCP-A -> GCP-B -> 111 local; Gemini is owned by the caller/AI Router as the
+final non-Ollama fallback.
 """
 
 from __future__ import annotations
@@ -31,28 +31,6 @@ OllamaWorkloadType = Literal[
     "dr",
 ]
 
-_GCP_A_PREFERRED_WORKLOADS = {
-    "interactive",
-    "healthcheck",
-    "alert_fast",
-    "batch",
-    "embedding",
-    "rag",
-    "code_review",
-    "shadow",
-    "canary",
-    "deep_rca",
-    "image_analysis",
-    "hermes",
-}
-
-_LOCAL_PREFERRED_WORKLOADS = {
-    "local_required",
-    "privacy_sensitive",
-    "dr",
-}
-
-
 class _OllamaSettings(Protocol):
     OLLAMA_URL: str
     OLLAMA_SECONDARY_URL: str
@@ -73,56 +51,47 @@ def resolve_ollama_selection(
     config: _OllamaSettings | None = None,
 ) -> OllamaEndpointSelection:
     """Return the preferred Ollama endpoint for a workload class."""
+    return resolve_ollama_order(workload_type, config=config)[0]
+
+
+def resolve_ollama_order(
+    workload_type: OllamaWorkloadType = "interactive",
+    *,
+    config: _OllamaSettings | None = None,
+) -> tuple[OllamaEndpointSelection, ...]:
+    """Return the global Ollama fallback order: GCP-A -> GCP-B -> 111."""
     cfg = config or settings
-    primary = cfg.OLLAMA_URL
-    secondary = cfg.OLLAMA_SECONDARY_URL
-    fallback = cfg.OLLAMA_FALLBACK_URL
 
-    if workload_type in _LOCAL_PREFERRED_WORKLOADS:
-        if fallback:
-            return OllamaEndpointSelection(
-                url=fallback,
-                provider_name="ollama_local",
+    candidates = (
+        (cfg.OLLAMA_URL, "ollama_gcp_a", "global_primary_gcp_a"),
+        (cfg.OLLAMA_SECONDARY_URL, "ollama_gcp_b", "global_secondary_gcp_b"),
+        (cfg.OLLAMA_FALLBACK_URL, "ollama_local", "global_local_111"),
+    )
+    selections: list[OllamaEndpointSelection] = []
+    seen: set[str] = set()
+    for url, provider_name, reason in candidates:
+        if not url or url in seen:
+            continue
+        seen.add(url)
+        selections.append(
+            OllamaEndpointSelection(
+                url=url,
+                provider_name=provider_name,
                 workload_type=workload_type,
-                reason="local_heavy_or_privacy_lane",
+                reason=reason,
             )
-        if secondary:
-            return OllamaEndpointSelection(
-                url=secondary,
-                provider_name="ollama_gcp_b",
-                workload_type=workload_type,
-                reason="local_missing_gcp_b_fallback",
-            )
-
-    if workload_type not in _GCP_A_PREFERRED_WORKLOADS and secondary:
-        return OllamaEndpointSelection(
-            url=secondary,
-            provider_name="ollama_gcp_b",
-            workload_type=workload_type,
-            reason="gcp_b_default_non_alert_lane",
         )
 
-    if primary:
-        return OllamaEndpointSelection(
-            url=primary,
-            provider_name="ollama_gcp_a",
-            workload_type=workload_type,
-            reason="primary_interactive_lane",
-        )
+    if selections:
+        return tuple(selections)
 
-    if secondary:
-        return OllamaEndpointSelection(
-            url=secondary,
-            provider_name="ollama_gcp_b",
+    return (
+        OllamaEndpointSelection(
+            url="",
+            provider_name="ollama_unconfigured",
             workload_type=workload_type,
-            reason="primary_missing_gcp_b_fallback",
-        )
-
-    return OllamaEndpointSelection(
-        url=fallback,
-        provider_name="ollama_local",
-        workload_type=workload_type,
-        reason="gcp_missing_local_fallback",
+            reason="no_ollama_endpoint_configured",
+        ),
     )
 
 

apps/api/tests/test_local_code_review_cloud_fallback.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+from types import SimpleNamespace
 from typing import Any
 
 import httpx
@@ -32,6 +33,14 @@ async def _noop_save(*args: Any, **kwargs: Any) -> None:
     return None
 
 
+def _fake_ollama_order(_workload_type: str) -> tuple[SimpleNamespace, ...]:
+    return (
+        SimpleNamespace(url="http://gcp-a:11434", provider_name="ollama_gcp_a"),
+        SimpleNamespace(url="http://gcp-b:11434", provider_name="ollama_gcp_b"),
+        SimpleNamespace(url="http://local-111:11434", provider_name="ollama_local"),
+    )
+
+
 @pytest.mark.asyncio
 async def test_large_pr_uses_local_ollama_when_gemini_fallback_disabled(
     monkeypatch: pytest.MonkeyPatch,
@@ -43,8 +52,8 @@ async def test_large_pr_uses_local_ollama_when_gemini_fallback_disabled(
     )
     monkeypatch.setattr(
         review_module,
-        "resolve_ollama_endpoint",
-        lambda workload_type: "http://local-111:11434",
+        "resolve_ollama_order",
+        _fake_ollama_order,
     )
 
     client = _FakeClient()
@@ -68,8 +77,8 @@ async def test_large_pr_uses_local_ollama_when_gemini_fallback_disabled(
     )
 
     assert result is not None
-    assert result["provider"] == "ollama"
-    assert client.posted_urls == ["http://local-111:11434/api/generate"]
+    assert result["provider"] == "ollama_gcp_a"
+    assert client.posted_urls == ["http://gcp-a:11434/api/generate"]
 
 
 @pytest.mark.asyncio
@@ -83,8 +92,8 @@ async def test_ollama_failure_does_not_fall_back_to_gemini_by_default(
     )
     monkeypatch.setattr(
         review_module,
-        "resolve_ollama_endpoint",
-        lambda workload_type: "http://local-111:11434",
+        "resolve_ollama_order",
+        _fake_ollama_order,
     )
 
     client = _FakeClient(fail=True)
@@ -110,7 +119,11 @@ async def test_ollama_failure_does_not_fall_back_to_gemini_by_default(
     assert result is not None
     assert result["provider"] == "ollama_unavailable"
     assert result["cloud_fallback_skipped"] is True
-    assert client.posted_urls == ["http://local-111:11434/api/generate"]
+    assert client.posted_urls == [
+        "http://gcp-a:11434/api/generate",
+        "http://gcp-b:11434/api/generate",
+        "http://local-111:11434/api/generate",
+    ]
 
 
 @pytest.mark.asyncio

apps/api/tests/test_ollama_endpoint_resolver.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 from types import SimpleNamespace
 
 from src.services.ollama_endpoint_resolver import (
+    resolve_ollama_order,
     resolve_ollama_selection,
 )
 
@@ -20,10 +21,13 @@ def _settings(
     )
 
 
-def test_non_sensitive_workloads_prefer_gcp_a_lane() -> None:
+def test_all_workloads_prefer_gcp_a_lane() -> None:
     cfg = _settings()
 
     for workload in (
+        "interactive",
+        "healthcheck",
+        "alert_fast",
         "batch",
         "embedding",
         "rag",
@@ -33,29 +37,31 @@ def test_non_sensitive_workloads_prefer_gcp_a_lane() -> None:
         "deep_rca",
         "image_analysis",
         "hermes",
+        "local_required",
+        "privacy_sensitive",
+        "dr",
     ):
         selection = resolve_ollama_selection(workload, config=cfg)
         assert selection.url == "http://192.168.0.110:11435"
         assert selection.provider_name == "ollama_gcp_a"
-        assert selection.reason == "primary_interactive_lane"
+        assert selection.reason == "global_primary_gcp_a"
 
 
-def test_interactive_workloads_stay_on_gcp_a() -> None:
+def test_all_workloads_share_global_ollama_order() -> None:
     cfg = _settings()
 
-    for workload in ("interactive", "healthcheck", "alert_fast"):
-        selection = resolve_ollama_selection(workload, config=cfg)
-        assert selection.url == "http://192.168.0.110:11435"
-        assert selection.provider_name == "ollama_gcp_a"
-
-
-def test_local_required_workloads_use_local_lane() -> None:
-    cfg = _settings()
-
-    for workload in ("local_required", "privacy_sensitive", "dr"):
-        selection = resolve_ollama_selection(workload, config=cfg)
-        assert selection.url == "http://192.168.0.110:11437"
-        assert selection.provider_name == "ollama_local"
+    for workload in ("interactive", "deep_rca", "local_required", "privacy_sensitive", "dr"):
+        order = resolve_ollama_order(workload, config=cfg)
+        assert [selection.url for selection in order] == [
+            "http://192.168.0.110:11435",
+            "http://192.168.0.110:11436",
+            "http://192.168.0.110:11437",
+        ]
+        assert [selection.provider_name for selection in order] == [
+            "ollama_gcp_a",
+            "ollama_gcp_b",
+            "ollama_local",
+        ]
 
 
 def test_non_sensitive_workloads_fall_back_to_gcp_b_when_primary_missing() -> None:
@@ -64,4 +70,10 @@ def test_non_sensitive_workloads_fall_back_to_gcp_b_when_primary_missing() -> No
     selection = resolve_ollama_selection("embedding", config=cfg)
     assert selection.url == "http://192.168.0.110:11436"
     assert selection.provider_name == "ollama_gcp_b"
-    assert selection.reason == "primary_missing_gcp_b_fallback"
+    assert selection.reason == "global_secondary_gcp_b"
+
+    order = resolve_ollama_order("embedding", config=cfg)
+    assert [selection.provider_name for selection in order] == [
+        "ollama_gcp_b",
+        "ollama_local",
+    ]