CMR-11195: remove /health caching, updates readme

Author: daniel-zamora

search-proxy/README.md

@@ -89,7 +89,7 @@ Always returns HTTP 200. Used for ALB/ECS target group health checks so that Red
 
 ### `GET /health`
 
-Informational health check. Always returns HTTP 200 — dependencies report their status but do not affect the top-level `ok?`. The result is cached for 5 seconds (unhealthy results are not cached so recovery is visible immediately).
+Informational health check, not cached. Nothing automated polls it — ALB/ECS use `/health/shallow`. Currently always returns HTTP 200: dependencies report their status but do not affect the top-level `ok?`.
 
 ```json
 {

search-proxy/src/proxy/app.py

@@ -212,11 +212,6 @@ async def forward_to_backend(
     return response
 
 
-# Cached health check result with TTL-based expiration
-_health_cache: dict = {"result": None, "expires": 0.0}
-_HEALTH_CACHE_TTL = 5.0
-
-
 @app.get("/health/shallow")
 async def health_shallow():
     return JSONResponse(status_code=200, content={"ok?": True})
@@ -226,19 +221,9 @@ async def health_shallow():
 async def health(request: Request):
     """Health check matching CMR's {:ok? bool :dependencies {...}} format.
 
-    Each dependency reports ok? and optionally a problem string. Lane
-    status is included so the health endpoint doubles as the single
-    place to check lane utilization."""
-    now = time.monotonic()
-
-    # Return cached result if still valid
-    if _health_cache["result"] and now < _health_cache["expires"]:
-        cached = _health_cache["result"]
-        return JSONResponse(
-            status_code=cached["status_code"],
-            content=cached["content"],
-        )
-
+    Informational only and not cached — nothing automated polls this
+    (ALB/ECS use /health/shallow). It exists so an operator can see
+    dependency and lane-utilization status in one place."""
     dependencies = {}
 
     # Redis
@@ -286,12 +271,6 @@ async def health(request: Request):
     status_code = 200 if ok else 503
     content = {"ok?": ok, "dependencies": dependencies}
 
-    # Only cache healthy results so recovery is visible on the next check.
-    # Use monotonic time (consistent with the check at the top of this function).
-    if status_code == 200:
-        _health_cache["result"] = {"status_code": status_code, "content": content}
-        _health_cache["expires"] = now + _HEALTH_CACHE_TTL
-
     return JSONResponse(status_code=status_code, content=content)
 
 

search-proxy/test/test_app.py

@@ -7,7 +7,7 @@
 import httpx
 import pytest
 
-from proxy.app import DEFAULT_TOGGLES, _health_cache, app, filter_hop_headers
+from proxy.app import DEFAULT_TOGGLES, app, filter_hop_headers
 from proxy.cache import ResponseCache
 from proxy.config import LaneConfig, LanesConfig, ProxySettings
 from proxy.lanes import RequestLanes
@@ -80,9 +80,6 @@ async def client():
     app.state.backend.get = AsyncMock(return_value=make_backend_response())
     app.state.backend.request = AsyncMock(return_value=make_backend_response())
 
-    _health_cache["result"] = None
-    _health_cache["expires"] = 0.0
-
     async with httpx.AsyncClient(
         transport=httpx.ASGITransport(app=app),
         base_url="http://test",
@@ -131,8 +128,6 @@ async def test_health_reports_lane_at_capacity(self, client):
         fake_redis = app.state.redis
         future = time.time() + 300
         await fake_redis.zadd("lane:heavy:active", {f"req-{i}": future for i in range(50)})
-        _health_cache["result"] = None
-        _health_cache["expires"] = 0.0
         resp = await client.get("/health")
         data = resp.json()
         assert resp.status_code == 200
@@ -146,23 +141,10 @@ async def test_health_shallow_always_200(self, client):
         assert resp.status_code == 200
         assert resp.json()["ok?"] is True
 
-    async def test_health_caches_result(self, client):
-        """Rapid /health calls should hit the cache, not backend each time."""
-        app.state.backend.get.return_value = make_backend_response()
-        await client.get("/health")
-        await client.get("/health")
-        health_calls = [
-            c
-            for c in app.state.backend.get.call_args_list
-            if "/search/health" in str(c)
-        ]
-        assert len(health_calls) == 1
-
-    async def test_health_cache_expires(self, client):
-        """After TTL expires, /health should re-check the backend."""
+    async def test_health_not_cached(self, client):
+        """/health is not cached — every call re-checks the backend."""
         app.state.backend.get.return_value = make_backend_response()
         await client.get("/health")
-        _health_cache["expires"] = 0.0
         await client.get("/health")
         health_calls = [
             c