fix(rate-limit): normalize queue refresh settings (#5499)

Co-authored-by: KooshaPari <koosha@example.com>

open-sse/services/rateLimitManager.ts

@@ -140,9 +140,7 @@ function resolveMinTime(override: number | undefined | null): number {
 
 // Resolve a maxConcurrent override. 0 or missing means "effectively infinite".
 function resolveMaxConcurrent(override: number | undefined | null): number {
-  return typeof override === "number" && override > 0
-    ? override
-    : EFFECTIVELY_INFINITE_CONCURRENCY;
+  return typeof override === "number" && override > 0 ? override : EFFECTIVELY_INFINITE_CONCURRENCY;
 }
 
 function buildLimiterDefaults() {
@@ -159,14 +157,9 @@ function buildLimiterDefaults() {
 }
 
 function updateAllLimiterSettings() {
+  const defaults = buildLimiterDefaults();
   for (const limiter of limiters.values()) {
-    limiter.updateSettings({
-      maxConcurrent: currentRequestQueueSettings.concurrentRequests,
-      minTime: currentRequestQueueSettings.minTimeBetweenRequestsMs,
-      reservoir: currentRequestQueueSettings.requestsPerMinute,
-      reservoirRefreshAmount: currentRequestQueueSettings.requestsPerMinute,
-      reservoirRefreshInterval: 60 * 1000,
-    });
+    limiter.updateSettings(defaults);
   }
 }
 
@@ -234,7 +227,9 @@ function watchdogTick() {
         limiters.delete(key);
         lastDispatchAt.delete(key);
         limiterLastUsed.delete(key);
-        logRateLimit(`🧹 [RATE-LIMIT] Evicting idle limiter: ${key} (inactive for ${Math.round((now - lastUsed) / 1000)}s)`);
+        logRateLimit(
+          `🧹 [RATE-LIMIT] Evicting idle limiter: ${key} (inactive for ${Math.round((now - lastUsed) / 1000)}s)`
+        );
         trackAsyncOperation(limiter.disconnect());
       }
     }
@@ -535,7 +530,12 @@ export async function withRateLimit(provider, connectionId, model, fn, signal =
 
   // Proactive sliding-window fallback for header-less providers with a declared cap
   // (Fase 8.2). No-op unless PROVIDER_DEFAULT_RATE_LIMITS has an entry for `provider`.
-  await awaitProviderDefaultSlot(provider, connectionId, signal, currentRequestQueueSettings.maxWaitMs);
+  await awaitProviderDefaultSlot(
+    provider,
+    connectionId,
+    signal,
+    currentRequestQueueSettings.maxWaitMs
+  );
 
   const limiter = getLimiter(provider, connectionId, model);
   const maxWaitMs = currentRequestQueueSettings.maxWaitMs;

tests/unit/limiter-lifecycle.test.ts

@@ -202,3 +202,35 @@ test("after 429 teardown, next withRateLimit must get a fresh limiter and succee
   );
   assert.equal(result, "post-429", "post-429 request must return its value");
 });
+
+test("request queue refresh treats zero limits as unbounded for existing limiters", async () => {
+  await flushBackgroundWork();
+
+  const provider = "openai";
+  const connectionId = "lifecycle-test-conn-d";
+
+  rateLimitManager.enableRateLimitProtection(connectionId);
+  assert.equal(
+    await rateLimitManager.withRateLimit(
+      provider,
+      connectionId,
+      null,
+      async () => "before-refresh"
+    ),
+    "before-refresh"
+  );
+
+  await rateLimitManager.applyRequestQueueSettings({
+    enabled: true,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 100,
+    requestsPerMinute: 0,
+    concurrentRequests: 0,
+    minTimeBetweenRequestsMs: 0,
+  });
+
+  assert.equal(
+    await rateLimitManager.withRateLimit(provider, connectionId, null, async () => "after-refresh"),
+    "after-refresh"
+  );
+});