fix: acknowledge pi maintenance barrier

This commit is contained in:
2026-08-04 19:32:05 +02:00
parent 0b9ad7f53f
commit 5b3ce93e31
11 changed files with 168 additions and 52 deletions
+20 -4
View File
@@ -15,7 +15,7 @@ import { settingsRoutes, effectiveSettings } from "./routes/settings.js";
import { createPiModelLister } from "./pi/list-models.js";
import { loadSettings, type Settings } from "./settings/settings-store.js";
import { ReadinessManager } from "./runtime/readiness-manager.js";
import { createMaintenanceGate } from "./runtime/maintenance-gate.js";
import { MaintenanceBarrier } from "./runtime/maintenance-gate.js";
import { WorkspaceRegistry } from "./workspaces/registry.js";
import { createProductionWorkspaceDiagnoser } from "./workspaces/diagnostics.js";
import { workspaceRoutes, type WorkspaceDiagnoser } from "./routes/workspaces.js";
@@ -33,8 +33,7 @@ export interface BuildAppDeps {
workspaceRegistry?: WorkspaceRegistry;
workspaceDiagnoser?: WorkspaceDiagnoser;
workspaceRuntimeSupport?: (workspace: WorkspaceDescriptor) => boolean;
/** Returns true while a host maintenance transaction is preventing new runtimes. */
maintenanceGate?: () => boolean;
maintenanceBarrier?: MaintenanceBarrier;
}
export function buildApp(config: AppConfig, deps?: BuildAppDeps): FastifyInstance {
@@ -97,12 +96,27 @@ export function buildApp(config: AppConfig, deps?: BuildAppDeps): FastifyInstanc
app.get("/health", async () => ({ status: "ok" }));
app.get("/health/dwh", async () => tht.dbPing());
app.get("/me", async (req) => getPrincipal(req));
const maintenanceBarrier = deps?.maintenanceBarrier ?? new MaintenanceBarrier();
sessionRoutes(app, {
mgr, tht: tht as ThtRunner, hub, getSettings, readiness, listModels, workspaceRegistry,
dwhPrecheck: config.dwhPrecheck,
legacyWorkspaceMode: config.legacyWorkspaceMode,
workspaceRuntimeSupport,
maintenanceGate: deps?.maintenanceGate ?? createMaintenanceGate(config.maintenanceFile),
maintenanceBarrier,
});
app.post("/internal/maintenance/activate", async (req, reply) => {
if (!isLoopback(req.ip) || req.principal?.subject !== "thothctl-maintenance") return reply.code(403).send({ error: "loopback maintenance control required" });
await maintenanceBarrier.activate();
return maintenanceBarrier.status();
});
app.post("/internal/maintenance/deactivate", async (req, reply) => {
if (!isLoopback(req.ip) || req.principal?.subject !== "thothctl-maintenance") return reply.code(403).send({ error: "loopback maintenance control required" });
maintenanceBarrier.deactivate();
return maintenanceBarrier.status();
});
app.get("/internal/maintenance/status", async (req, reply) => {
if (!isLoopback(req.ip) || req.principal?.subject !== "thothctl-maintenance") return reply.code(403).send({ error: "loopback maintenance control required" });
return maintenanceBarrier.status();
});
sqlRoutes(app, { tht: tht as ThtRunner, getSettings });
metaRoutes(app, { harnessDir: config.harnessDir, listModels });
@@ -111,3 +125,5 @@ export function buildApp(config: AppConfig, deps?: BuildAppDeps): FastifyInstanc
return app;
}
function isLoopback(ip: string): boolean { return ip === "127.0.0.1" || ip === "::1" || ip === "::ffff:127.0.0.1"; }
+10 -4
View File
@@ -9,6 +9,7 @@ import type { ReadinessManager } from "../runtime/readiness-manager.js";
import type { ListModelsFn } from "./meta.js";
import type { WorkspaceRegistry } from "../workspaces/registry.js";
import type { WorkspaceDescriptor } from "../workspaces/schema.js";
import type { MaintenanceBarrier } from "../runtime/maintenance-gate.js";
const BOOTSTRAP_FAILURE_MESSAGE =
"Session startup failed. Check configuration and connectivity, then Resume the session.";
@@ -37,8 +38,7 @@ export function sessionRoutes(
legacyWorkspaceMode?: boolean;
/** Fail-closed installation/runtime transport capability check. */
workspaceRuntimeSupport: (workspace: WorkspaceDescriptor) => boolean;
/** Host-controlled admission guard. Existing runtimes deliberately continue. */
maintenanceGate: () => boolean;
maintenanceBarrier: MaintenanceBarrier;
},
) {
const lifecycleTails = new Map<string, Promise<void>>();
@@ -81,6 +81,14 @@ export function sessionRoutes(
code: "maintenance",
error: "Session admission is temporarily paused for maintenance. Try again shortly.",
});
const admissionLeases = new WeakMap<object, () => void>();
app.addHook("preHandler", async (req, reply) => {
if (req.method !== "POST" || !(req.url === "/sessions" || /^\/sessions\/[^/]+\/resume(?:\?|$)/.test(req.url))) return;
const release = d.maintenanceBarrier.acquire();
if (!release) return maintenanceReply(reply);
admissionLeases.set(req, release);
});
app.addHook("onResponse", async (req) => { admissionLeases.get(req)?.(); });
/** Include retained historical descriptors so removed workspaces remain resumable. */
const sessionRevisions = async () => {
@@ -279,7 +287,6 @@ export function sessionRoutes(
});
app.post("/sessions", async (req, reply) => {
if (d.maintenanceGate()) return maintenanceReply(reply);
const b = req.body as {
question: string; name?: string; workspace?: string; workspaceId?: string;
provider?: string; model?: string; thinking?: string;
@@ -510,7 +517,6 @@ export function sessionRoutes(
return reply.code(204).send();
});
app.post("/sessions/:id/resume", async (req, reply) => {
if (d.maintenanceGate()) return maintenanceReply(reply);
const id = (req.params as any).id;
const principal = getPrincipal(req);
return withSessionLifecycle(id, async () => {
+25 -8
View File
@@ -1,10 +1,27 @@
import { existsSync } from "node:fs";
/** An in-process admission barrier. A lease spans the complete create/resume decision. */
export class MaintenanceBarrier {
private active = false;
private admissions = 0;
private waiters: (() => void)[] = [];
/**
* A host-side lifecycle transaction creates this marker before it checks for active sessions.
* The marker is intentionally only an admission gate: it must never terminate existing Pi
* processes or make their in-flight work unavailable.
*/
export function createMaintenanceGate(markerFile: string): () => boolean {
return () => existsSync(markerFile);
acquire(): (() => void) | undefined {
if (this.active) return undefined;
this.admissions += 1;
let released = false;
return () => {
if (released) return;
released = true;
this.admissions -= 1;
if (this.admissions === 0) this.waiters.splice(0).forEach((resolve) => resolve());
};
}
async activate(): Promise<void> {
this.active = true;
if (this.admissions === 0) return;
await new Promise<void>((resolve) => this.waiters.push(resolve));
}
deactivate(): void { this.active = false; }
status(): { active: boolean; admissions: number } { return { active: this.active, admissions: this.admissions }; }
}
+19
View File
@@ -0,0 +1,19 @@
import { test, expect } from "vitest";
import { MaintenanceBarrier } from "../src/runtime/maintenance-gate.js";
test("activation waits for an in-flight admission lease and rejects later admissions", async () => {
const gate = new MaintenanceBarrier();
const release = gate.acquire();
expect(release).toBeTypeOf("function");
let acknowledged = false;
const activation = gate.activate().then(() => { acknowledged = true; });
await Promise.resolve();
expect(acknowledged).toBe(false);
expect(gate.acquire()).toBeUndefined();
release?.();
await activation;
expect(acknowledged).toBe(true);
expect(gate.status()).toEqual({ active: true, admissions: 0 });
gate.deactivate();
expect(gate.acquire()).toBeTypeOf("function");
});
+6 -4
View File
@@ -7,6 +7,7 @@ import { tmpdir } from "node:os";
import { buildApp as buildRealApp } from "../src/app.js";
import { loadConfig } from "../src/config.js";
import { SseHub } from "../src/sse/sse-hub.js";
import { MaintenanceBarrier } from "../src/runtime/maintenance-gate.js";
const FAKE = path.resolve("../harness/tests/fake_pi/fake_pi_rpc.mjs");
const SCRIPT = path.resolve("../harness/tests/fake_pi/scripts/f1_disambiguation.json");
@@ -76,8 +77,10 @@ test("upstream requests without a principal fail before a Pi runtime can be crea
});
test("maintenance rejects new and resumed session admission without interrupting running sessions", async () => {
const maintenanceBarrier = new MaintenanceBarrier();
await maintenanceBarrier.activate();
const app = buildApp(loadConfig({ AUTH_MODE: "upstream", THT_HARNESS_DIR: "../harness" }), {
maintenanceGate: () => true,
maintenanceBarrier,
thtRunner: { withPrincipal: () => ({ sessionShow: async () => ({ id: "open", status: "open" }) }) } as any,
});
@@ -94,7 +97,7 @@ test("maintenance rejects new and resumed session admission without interrupting
expect(resume.json()).toEqual(create.json());
});
test("the on-disk maintenance marker gates admission in an upstream server profile", async () => {
test("a server-profile marker does not weaken the in-process maintenance gate", async () => {
const dir = mkdtempSync(path.join(tmpdir(), "tht-maintenance-"));
try {
const marker = path.join(dir, "maintenance.json");
@@ -105,8 +108,7 @@ test("the on-disk maintenance marker gates admission in an upstream server profi
const response = await app.inject({
method: "POST", url: "/sessions", headers: aliceHeaders, payload: { question: "q" },
});
expect(response.statusCode).toBe(503);
expect(response.json().code).toBe("maintenance");
expect(response.statusCode).not.toBe(503);
} finally {
rmSync(dir, { recursive: true, force: true });
}