fix(cloud-api): keep health live nonblocking (#1808)

Route /health/live through cached runtime summary so DB durable retry cannot block liveness probes. Keep /health as the full durable readiness endpoint.\n\nRefs #1807.

internal/cloud/server.ts

@@ -262,20 +262,24 @@ function ensurePreinstalledHwpodSpec({ workspace }) {
 export async function buildHealthPayload(options = {}) {
   const serviceId = CLOUD_API_SERVICE_ID;
   const env = options.env ?? process.env;
+  const healthMode = options.healthMode === "live" ? "live" : "readiness";
+  const liveProbe = healthMode === "live";
   ensureCodeAgentRuntimeBase(env);
   const metadata = buildMetadataFromEnv(env, {
     serviceId,
     fallbackImageRepository: "ghcr.io/pikastech/hwlab-cloud-api"
   });
-  const dbProbe = await buildDbRuntimeReadiness(env, options.dbProbe);
+  const runtimeStore = options.runtimeStore ?? createConfiguredCloudRuntimeStore({ ...options, env });
+  const dbProbe = await buildDbRuntimeReadiness(env, liveProbe ? { ...(options.dbProbe ?? {}), probe: false } : options.dbProbe);
   const codeAgent = await describeCodeAgentAvailability(env, options);
-  const runtime = await runtimeReadiness(options.runtimeStore ?? createConfiguredCloudRuntimeStore({ ...options, env }));
+  const runtime = liveProbe ? runtimeReadinessSnapshot(runtimeStore) : await runtimeReadiness(runtimeStore);
   const db = applyRuntimeDbReadinessLayers(dbProbe, runtime);
   const readiness = buildCloudApiReadiness({ db, codeAgent, runtime });
 
   return {
     serviceId,
     environment: runtimeEnvironment(env),
+    healthMode,
     status: readiness.status,
     ready: readiness.ready,
     revision: metadata.revision,
@@ -300,11 +304,54 @@ export async function buildHealthPayload(options = {}) {
   };
 }
 
+function runtimeReadinessSnapshot(runtimeStore) {
+  if (runtimeStore && typeof runtimeStore.summary === "function") {
+    try {
+      const summary = runtimeStore.summary();
+      return {
+        ...summary,
+        readinessFresh: false,
+        readinessSource: "cached-summary"
+      };
+    } catch (error) {
+      return {
+        adapter: "unknown",
+        durable: false,
+        ready: false,
+        status: "degraded",
+        blocker: "runtime_readiness_snapshot_failed",
+        reason: error instanceof Error ? error.message : "Runtime readiness snapshot failed",
+        readinessFresh: false,
+        readinessSource: "cached-summary",
+        valuesRedacted: true,
+        endpointRedacted: true
+      };
+    }
+  }
+  return {
+    adapter: "unknown",
+    durable: false,
+    ready: false,
+    status: "degraded",
+    blocker: "runtime_readiness_snapshot_unavailable",
+    reason: "Runtime store does not expose a cached summary for /health/live",
+    readinessFresh: false,
+    readinessSource: "cached-summary",
+    valuesRedacted: true,
+    endpointRedacted: true
+  };
+}
+
 async function routeRequest(request, response, options) {
   const url = new URL(request.url || "/", "http://hwlab-cloud-api.local");
 
-  if (request.method === "GET" && (url.pathname === "/health" || url.pathname === "/health/live")) {
-    sendJson(response, 200, await buildHealthPayload(options));
+  if (request.method === "GET" && url.pathname === "/health/live") {
+    sendJson(response, 200, await buildHealthPayload({ ...options, healthMode: "live" }));
+    return;
+  }
+
+  if (request.method === "GET" && url.pathname === "/health") {
+    sendJson(response, 200, await buildHealthPayload({ ...options, healthMode: "readiness" }));
     return;
   }