fix: serve a deterministic internal embedding fixture for P2 acceptance

backend/scripts/p2-acceptance.mjs

@@ -766,26 +766,32 @@ async function writeInstallationFiles(ctx) {
   ].join("\n") + "\n";
   await atomicWrite(operatorEnvPath, operatorEnv);
   await atomicWrite(join(ctx.run.root, "installation", "pi-auth.json"), JSON.stringify({ fixture: true }));
+  const embeddingStubPath = join(installationDir, "embedding-stub.py");
+  await atomicWrite(embeddingStubPath, EMBEDDING_STUB_SOURCE);
   const override = {
     services: {
       core: {
         image: ctx.coreImageTag,
-        environment: {
-          THT_INTERNAL_EMBEDDING_URL: `http://host.docker.internal:${ctx.fixturePorts.embedding}`,
-        },
         extra_hosts: ["host.docker.internal:host-gateway"],
       },
       "workspace-maintenance": {
         image: ctx.coreImageTag,
-        environment: {
-          THT_INTERNAL_EMBEDDING_URL: `http://host.docker.internal:${ctx.fixturePorts.embedding}`,
-        },
         extra_hosts: ["host.docker.internal:host-gateway"],
       },
       qdrant: {
         ports: [`127.0.0.1:${qdrantPort}:6333`],
         restart: "no",
       },
+      // Deterministic Ollama-compatible embedding fixture on the internal allowlisted host
+      // name `embedding` (http://embedding:11434). Replaces the real Ollama service entirely.
+      embedding: {
+        image: ctx.coreImageTag,
+        entrypoint: ["python3", "/stub.py"],
+        volumes: [
+          { type: "bind", source: embeddingStubPath, target: "/stub.py", read_only: true },
+        ],
+        healthcheck: { disable: true },
+      },
     },
   };
   await atomicWrite(fixtureOverridePath, yamlStringify(override, { lineWidth: 0, sortMapEntries: false }));
@@ -888,7 +894,7 @@ async function dockerCompose(ctx, commandArgs, { allowFailure = false, maxOutput
 }
 
 async function startQdrant(ctx) {
-  await dockerCompose(ctx, ["up", "-d", "qdrant"]);
+  await dockerCompose(ctx, ["up", "-d", "qdrant", "embedding"]);
   for (let attempt = 0; attempt < 60; attempt += 1) {
     try {
       const response = await fetch(`http://127.0.0.1:${ctx.fixturePorts.qdrant}/collections`);
@@ -972,6 +978,30 @@ function sameSet(left, right) {
   return JSON.stringify([...left].sort()) === JSON.stringify([...right].sort());
 }
 
+const EMBEDDING_STUB_SOURCE = String.raw`import json
+from http.server import BaseHTTPRequestHandler, HTTPServer
+
+class _Handler(BaseHTTPRequestHandler):
+    def do_POST(self):
+        length = int(self.headers.get("Content-Length", "0"))
+        payload = json.loads(self.rfile.read(length))
+        inputs = payload.get("input", [])
+        if isinstance(inputs, str):
+            inputs = [inputs]
+        embeddings = [[0.01] * 1024 for _ in inputs]
+        body = json.dumps({"model": payload.get("model", "qwen3-embedding:0.6b"), "embeddings": embeddings}).encode("utf-8")
+        self.send_response(200)
+        self.send_header("Content-Type", "application/json")
+        self.send_header("Content-Length", str(len(body)))
+        self.end_headers()
+        self.wfile.write(body)
+
+    def log_message(self, *args):
+        pass
+
+HTTPServer(("0.0.0.0", 11434), _Handler).serve_forever()
+`;
+
 function realUserHome() {
   try {
     const output = execFileSync("bash", ["-lc", 'printf "%s" ~'], { encoding: "utf8" }).trim();