refactor(gateway): consolidate lifecycle lazy boundary (#74105)

* refactor(gateway): consolidate lifecycle lazy boundary

* test(gateway): cover quoted lifecycle imports

CHANGELOG.md

@@ -21,6 +21,7 @@ Docs: https://docs.openclaw.ai
 
 ### Fixes
 
+- Build/Gateway: route restart, shutdown, respawn, diagnostics, command-queue cleanup, and runtime cleanup through one stable gateway lifecycle runtime entry so rebuilt packages do not strand long-running gateways on stale hashed chunks. Carries forward #73964. Thanks @pashpashpash.
 - Agents/errors: suppress malformed streaming tool-call JSON fragments before they reach chat surfaces while preserving provider request-validation diagnostics. Fixes #59076; keeps #59080 as duplicate coverage. (#59118) Thanks @singleGanghood.
 - CLI/models: restore provider-filtered `models list --all --provider <id>` rows for providers without manifest/static catalog coverage, including Anthropic and Amazon Bedrock, while keeping the compatibility fallback off expensive availability and resolver paths. Thanks @shakkernerd.
 - CLI/models: move the OpenAI listable catalog into the plugin manifest so `models list --all --provider openai` uses the manifest fast path instead of loading provider runtime normalization hooks. Thanks @shakkernerd.

src/cli/gateway-cli/lifecycle.runtime.ts

@@ -0,0 +1,34 @@
+export {
+  abortEmbeddedPiRun,
+  getActiveEmbeddedRunCount,
+  waitForActiveEmbeddedRuns,
+} from "../../agents/pi-embedded-runner/runs.js";
+export { getRuntimeConfig } from "../../config/config.js";
+export {
+  respawnGatewayProcessForUpdate,
+  restartGatewayProcessWithFreshPid,
+} from "../../infra/process-respawn.js";
+export {
+  resolveGatewayRestartDeferralTimeoutMs,
+  consumeGatewayRestartIntentSync,
+  consumeGatewaySigusr1RestartAuthorization,
+  isGatewaySigusr1RestartExternallyAllowed,
+  markGatewaySigusr1RestartHandled,
+  peekGatewaySigusr1RestartReason,
+  resetGatewayRestartStateForInProcessRestart,
+  scheduleGatewaySigusr1Restart,
+} from "../../infra/restart.js";
+export { markUpdateRestartSentinelFailure } from "../../infra/restart-sentinel.js";
+export { detectRespawnSupervisor } from "../../infra/supervisor-markers.js";
+export { writeDiagnosticStabilityBundleForFailureSync } from "../../logging/diagnostic-stability-bundle.js";
+export {
+  getActiveBundledRuntimeDepsInstallCount,
+  waitForBundledRuntimeDepsInstallIdle,
+} from "../../plugins/bundled-runtime-deps-activity.js";
+export {
+  getActiveTaskCount,
+  markGatewayDraining,
+  resetAllLanes,
+  waitForActiveTasks,
+} from "../../process/command-queue.js";
+export { reloadTaskRegistryFromStore } from "../../tasks/runtime-internal.js";

src/cli/gateway-cli/run-loop.ts

@@ -15,48 +15,12 @@ const UPDATE_RESPAWN_HEALTH_POLL_MS = 200;
 type GatewayRunSignalAction = "stop" | "restart";
 type RestartDrainTimeoutMs = number | undefined;
 
-type EmbeddedRunsModule = typeof import("../../agents/pi-embedded-runner/runs.js");
-type RuntimeConfigModule = typeof import("../../config/config.js");
-type ProcessRespawnModule = typeof import("../../infra/process-respawn.js");
-type RestartSentinelModule = typeof import("../../infra/restart-sentinel.js");
-type RestartModule = typeof import("../../infra/restart.js");
-type SupervisorMarkersModule = typeof import("../../infra/supervisor-markers.js");
-type DiagnosticStabilityBundleModule =
-  typeof import("../../logging/diagnostic-stability-bundle.js");
-type BundledRuntimeDepsActivityModule =
-  typeof import("../../plugins/bundled-runtime-deps-activity.js");
-type CommandQueueModule = typeof import("../../process/command-queue.js");
-type RuntimeInternalModule = typeof import("../../tasks/runtime-internal.js");
+type GatewayLifecycleRuntimeModule = typeof import("./lifecycle.runtime.js");
 
-let embeddedRunsModule: Promise<EmbeddedRunsModule> | undefined;
-let runtimeConfigModule: Promise<RuntimeConfigModule> | undefined;
-let processRespawnModule: Promise<ProcessRespawnModule> | undefined;
-let restartSentinelModule: Promise<RestartSentinelModule> | undefined;
-let restartModule: Promise<RestartModule> | undefined;
-let supervisorMarkersModule: Promise<SupervisorMarkersModule> | undefined;
-let diagnosticStabilityBundleModule: Promise<DiagnosticStabilityBundleModule> | undefined;
-let bundledRuntimeDepsActivityModule: Promise<BundledRuntimeDepsActivityModule> | undefined;
-let commandQueueModule: Promise<CommandQueueModule> | undefined;
-let runtimeInternalModule: Promise<RuntimeInternalModule> | undefined;
+let gatewayLifecycleRuntimeModule: Promise<GatewayLifecycleRuntimeModule> | undefined;
 
-const loadEmbeddedRunsModule = () =>
-  (embeddedRunsModule ??= import("../../agents/pi-embedded-runner/runs.js"));
-const loadRuntimeConfigModule = () => (runtimeConfigModule ??= import("../../config/config.js"));
-const loadProcessRespawnModule = () =>
-  (processRespawnModule ??= import("../../infra/process-respawn.js"));
-const loadRestartSentinelModule = () =>
-  (restartSentinelModule ??= import("../../infra/restart-sentinel.js"));
-const loadRestartModule = () => (restartModule ??= import("../../infra/restart.js"));
-const loadSupervisorMarkersModule = () =>
-  (supervisorMarkersModule ??= import("../../infra/supervisor-markers.js"));
-const loadDiagnosticStabilityBundleModule = () =>
-  (diagnosticStabilityBundleModule ??= import("../../logging/diagnostic-stability-bundle.js"));
-const loadBundledRuntimeDepsActivityModule = () =>
-  (bundledRuntimeDepsActivityModule ??= import("../../plugins/bundled-runtime-deps-activity.js"));
-const loadCommandQueueModule = () =>
-  (commandQueueModule ??= import("../../process/command-queue.js"));
-const loadRuntimeInternalModule = () =>
-  (runtimeInternalModule ??= import("../../tasks/runtime-internal.js"));
+const loadGatewayLifecycleRuntimeModule = () =>
+  (gatewayLifecycleRuntimeModule ??= import("./lifecycle.runtime.js"));
 
 function createRestartIterationHook(onRestart: () => Promise<void> | void): () => Promise<boolean> {
   let isFirstIteration = true;
@@ -137,7 +101,7 @@ export async function runGatewayLoop(params: {
   };
   const writeStabilityBundle = async (reason: string, error?: unknown) => {
     const { writeDiagnosticStabilityBundleForFailureSync } =
-      await loadDiagnosticStabilityBundleModule();
+      await loadGatewayLifecycleRuntimeModule();
     const result = writeDiagnosticStabilityBundleForFailureSync(reason, error);
     if ("message" in result) {
       gatewayLog.warn(result.message);
@@ -165,10 +129,12 @@ export async function runGatewayLoop(params: {
   const handleRestartAfterServerClose = async (restartReason?: string) => {
     const hadLock = await releaseLockIfHeld();
     const isUpdateRestart = restartReason === "update.run";
-    const { respawnGatewayProcessForUpdate, restartGatewayProcessWithFreshPid } =
-      await loadProcessRespawnModule();
-    const { detectRespawnSupervisor } = await loadSupervisorMarkersModule();
-    const { markUpdateRestartSentinelFailure } = await loadRestartSentinelModule();
+    const {
+      detectRespawnSupervisor,
+      markUpdateRestartSentinelFailure,
+      respawnGatewayProcessForUpdate,
+      restartGatewayProcessWithFreshPid,
+    } = await loadGatewayLifecycleRuntimeModule();
 
     if (isUpdateRestart) {
       const respawn = respawnGatewayProcessForUpdate();
@@ -279,10 +245,8 @@ export async function runGatewayLoop(params: {
   const SHUTDOWN_TIMEOUT_MS = SUPERVISOR_STOP_TIMEOUT_MS - 5_000;
   const resolveRestartDrainTimeoutMs = async (): Promise<RestartDrainTimeoutMs> => {
     try {
-      const [{ getRuntimeConfig }, { resolveGatewayRestartDeferralTimeoutMs }] = await Promise.all([
-        loadRuntimeConfigModule(),
-        loadRestartModule(),
-      ]);
+      const { getRuntimeConfig, resolveGatewayRestartDeferralTimeoutMs } =
+        await loadGatewayLifecycleRuntimeModule();
       const timeoutMs = getRuntimeConfig().gateway?.reload?.deferralTimeoutMs;
       return resolveGatewayRestartDeferralTimeoutMs(timeoutMs);
     } catch {
@@ -351,15 +315,16 @@ export async function runGatewayLoop(params: {
         // On restart, wait for in-flight agent turns to finish before
         // tearing down the server so buffered messages are delivered.
         if (isRestart) {
-          const [
-            { abortEmbeddedPiRun, getActiveEmbeddedRunCount, waitForActiveEmbeddedRuns },
-            { getActiveBundledRuntimeDepsInstallCount, waitForBundledRuntimeDepsInstallIdle },
-            { getActiveTaskCount, markGatewayDraining, waitForActiveTasks },
-          ] = await Promise.all([
-            loadEmbeddedRunsModule(),
-            loadBundledRuntimeDepsActivityModule(),
-            loadCommandQueueModule(),
-          ]);
+          const {
+            abortEmbeddedPiRun,
+            getActiveBundledRuntimeDepsInstallCount,
+            getActiveEmbeddedRunCount,
+            getActiveTaskCount,
+            markGatewayDraining,
+            waitForActiveEmbeddedRuns,
+            waitForActiveTasks,
+            waitForBundledRuntimeDepsInstallIdle,
+          } = await loadGatewayLifecycleRuntimeModule();
           const createStillPendingDrainLogger = () =>
             setInterval(() => {
               gatewayLog.warn(
@@ -429,7 +394,7 @@ export async function runGatewayLoop(params: {
   const onSigterm = () => {
     gatewayLog.info("signal SIGTERM received");
     void (async () => {
-      const { consumeGatewayRestartIntentSync } = await loadRestartModule();
+      const { consumeGatewayRestartIntentSync } = await loadGatewayLifecycleRuntimeModule();
       request(consumeGatewayRestartIntentSync() ? "restart" : "stop", "SIGTERM");
     })();
   };
@@ -446,7 +411,7 @@ export async function runGatewayLoop(params: {
         markGatewaySigusr1RestartHandled,
         peekGatewaySigusr1RestartReason,
         scheduleGatewaySigusr1Restart,
-      } = await loadRestartModule();
+      } = await loadGatewayLifecycleRuntimeModule();
       const authorized = consumeGatewaySigusr1RestartAuthorization();
       if (!authorized) {
         if (!isGatewaySigusr1RestartExternallyAllowed()) {
@@ -482,15 +447,11 @@ export async function runGatewayLoop(params: {
       // new work from draining. The same boundary also discards stale restart
       // deferral timers and reloads the task registry from durable state so
       // cancelled/completed work is not kept alive by old in-memory maps.
-      const [
-        { resetAllLanes },
-        { resetGatewayRestartStateForInProcessRestart },
-        { reloadTaskRegistryFromStore },
-      ] = await Promise.all([
-        loadCommandQueueModule(),
-        loadRestartModule(),
-        loadRuntimeInternalModule(),
-      ]);
+      const {
+        reloadTaskRegistryFromStore,
+        resetAllLanes,
+        resetGatewayRestartStateForInProcessRestart,
+      } = await loadGatewayLifecycleRuntimeModule();
       resetAllLanes();
       resetGatewayRestartStateForInProcessRestart();
       reloadTaskRegistryFromStore();

src/infra/tsdown-config.test.ts

@@ -1,3 +1,4 @@
+import { readFileSync } from "node:fs";
 import { bundledPluginRoot } from "openclaw/plugin-sdk/test-fixtures";
 import { describe, expect, it } from "vitest";
 import tsdownConfig from "../../tsdown.config.ts";
@@ -41,6 +42,13 @@ function entryKeys(config: TsdownConfigEntry): string[] {
   return Object.keys(config.entry);
 }
 
+function entrySources(config: TsdownConfigEntry): Record<string, string> {
+  if (!config.entry || Array.isArray(config.entry)) {
+    return {};
+  }
+  return config.entry;
+}
+
 function hasBundledPluginRuntimeEntry(config: TsdownConfigEntry): boolean {
   const keys = entryKeys(config);
   return keys.includes("index") || keys.includes("runtime-api");
@@ -56,6 +64,10 @@ function unifiedDistGraph(): TsdownConfigEntry | undefined {
   );
 }
 
+function readGatewayRunLoopSource(): string {
+  return readFileSync(new URL("../cli/gateway-cli/run-loop.ts", import.meta.url), "utf8");
+}
+
 describe("tsdown config", () => {
   it("keeps core, plugin runtime, plugin-sdk, bundled root plugins, and bundled hooks in one dist graph", () => {
     const distGraph = unifiedDistGraph();
@@ -66,6 +78,7 @@ describe("tsdown config", () => {
         "agents/auth-profiles.runtime",
         "agents/model-catalog.runtime",
         "agents/models-config.runtime",
+        "cli/gateway-lifecycle.runtime",
         "plugins/memory-state",
         "subagent-registry.runtime",
         "task-registry-control.runtime",
@@ -85,6 +98,24 @@ describe("tsdown config", () => {
     );
   });
 
+  it("keeps gateway lifecycle lazy runtime behind one stable dist entry", () => {
+    const distGraph = unifiedDistGraph();
+
+    expect(entrySources(distGraph as TsdownConfigEntry)).toEqual(
+      expect.objectContaining({
+        "cli/gateway-lifecycle.runtime": "src/cli/gateway-cli/lifecycle.runtime.ts",
+      }),
+    );
+  });
+
+  it("routes gateway run-loop lifecycle imports through the stable runtime boundary", () => {
+    const importSpecifiers = [
+      ...readGatewayRunLoopSource().matchAll(/import\(["']([^"']+)["']\)/gu),
+    ].map((match) => match[1]);
+
+    expect(new Set(importSpecifiers)).toEqual(new Set(["./lifecycle.runtime.js"]));
+  });
+
   it("emits staged bundled plugins as separate extension graphs", () => {
     const stagedGraphs = asConfigArray(tsdownConfig).filter(
       (config) => typeof config.outDir === "string" && config.outDir.startsWith("dist/extensions/"),

tsdown.config.ts

@@ -212,6 +212,7 @@ function buildCoreDistEntries(): Record<string, string> {
     "agents/auth-profiles.runtime": "src/agents/auth-profiles.runtime.ts",
     "agents/model-catalog.runtime": "src/agents/model-catalog.runtime.ts",
     "agents/models-config.runtime": "src/agents/models-config.runtime.ts",
+    "cli/gateway-lifecycle.runtime": "src/cli/gateway-cli/lifecycle.runtime.ts",
     "plugins/memory-state": "src/plugins/memory-state.ts",
     "subagent-registry.runtime": "src/agents/subagent-registry.runtime.ts",
     "task-registry-control.runtime": "src/tasks/task-registry-control.runtime.ts",