fix(python,nodejs): serialize concurrent client startup

nodejs/src/client.ts

@@ -448,6 +448,8 @@ export class CopilotClient {
     private runtimePort: number | null = null;
     private actualHost: string = "localhost";
     private state: "disconnected" | "connecting" | "connected" | "error" = "disconnected";
+    /** Shared in-flight start; concurrent callers await it instead of spawning another CLI. */
+    private startPromise: Promise<void> | null = null;
     private sessions: Map<string, CopilotSession> = new Map();
     private stderrBuffer: string = ""; // Captures CLI stderr for error messages
     /** Resolved connection mode chosen in the constructor. */
@@ -935,6 +937,20 @@ export class CopilotClient {
             return;
         }
 
+        // Concurrent callers share one in-progress start instead of each spawning a CLI.
+        if (this.startPromise) {
+            return this.startPromise;
+        }
+
+        this.startPromise = this.doStart();
+        try {
+            await this.startPromise;
+        } finally {
+            this.startPromise = null;
+        }
+    }
+
+    private async doStart(): Promise<void> {
         this.forceStopping = false;
         this.connectionClosed = false;
         this.processTransportError = null;

nodejs/test/client.test.ts

@@ -62,6 +62,64 @@ describe("approveAll", () => {
 });
 
 describe("CopilotClient", () => {
+    it("start() is single-flight: concurrent callers share one startup", async () => {
+        const client = new CopilotClient({ autoStart: false });
+        onTestFinished(() => client.forceStop());
+
+        // Stub the underlying startup (doStart) that the single-flight guard
+        // dedupes. Transport-independent: this is the same regardless of the
+        // stdio vs in-process connection path. The delay makes all three
+        // start() calls overlap; on success it marks the client connected like
+        // the real doStart does.
+        const doStart = vi.fn().mockImplementation(
+            () =>
+                new Promise<void>((resolve) =>
+                    setTimeout(() => {
+                        (client as any).state = "connected";
+                        resolve();
+                    }, 50)
+                )
+        );
+        (client as any).doStart = doStart;
+
+        // Before the fix, each concurrent caller ran startup (and spawned its own
+        // CLI, orphaning all but the last). With single-flight they share one.
+        await Promise.all([client.start(), client.start(), client.start()]);
+
+        expect(doStart).toHaveBeenCalledTimes(1);
+        expect((client as any).state).toBe("connected");
+
+        // Once connected, a further start() is a no-op (no extra startup).
+        await client.start();
+        expect(doStart).toHaveBeenCalledTimes(1);
+    });
+
+    it("start() retries after a failed attempt (single-flight guard is cleared)", async () => {
+        const client = new CopilotClient({ autoStart: false });
+        onTestFinished(() => client.forceStop());
+
+        // Stub the underlying startup: fail once, then succeed. Transport-
+        // independent (does not depend on the stdio vs in-process path).
+        const doStart = vi
+            .fn()
+            .mockImplementationOnce(async () => {
+                (client as any).state = "error";
+                throw new Error("boom");
+            })
+            .mockImplementationOnce(async () => {
+                (client as any).state = "connected";
+            });
+        (client as any).doStart = doStart;
+
+        await expect(client.start()).rejects.toThrow(/boom/);
+        expect((client as any).state).toBe("error");
+
+        // The guard must have cleared so a later start() can retry.
+        await client.start();
+        expect(doStart).toHaveBeenCalledTimes(2);
+        expect((client as any).state).toBe("connected");
+    });
+
     it.each([
         {
             source: "connection path",

nodejs/test/e2e/session.e2e.test.ts

@@ -103,6 +103,10 @@ describe("Sessions", () => {
             workingDirectory: workDir,
             env,
             connection: RuntimeConnection.forStdio({ path: process.env.COPILOT_CLI_PATH }),
+            // Explicit token (matches createClient()/other passing resume tests): without it,
+            // useLoggedInUser defaults to true and the runtime falls back to ambient env-var
+            // auto-detection for the model call, which flakes on some hosts (e.g. Alpine ARM64).
+            gitHubToken: isCI ? DEFAULT_GITHUB_TOKEN : undefined,
         });
         onTestFinished(async () => {
             try {
@@ -127,6 +131,7 @@ describe("Sessions", () => {
             workingDirectory: workDir,
             env,
             connection: RuntimeConnection.forStdio({ path: process.env.COPILOT_CLI_PATH }),
+            gitHubToken: isCI ? DEFAULT_GITHUB_TOKEN : undefined,
         });
         onTestFinished(async () => {
             try {

python/copilot/client.py

@@ -1753,6 +1753,7 @@ def __init__(
         self._cli_process: subprocess.Popen | None = None
         self._client: JsonRpcClient | None = None
         self._state: _ConnectionState = "disconnected"
+        self._start_lock = asyncio.Lock()
         self._sessions: dict[str, CopilotSession] = {}
         self._sessions_lock = threading.Lock()
         self._github_token_providers: dict[str, _GitHubTokenProviderRegistration] = {}
@@ -1938,6 +1939,13 @@ async def start(self) -> None:
             >>> await client.start()
             >>> # Now ready to create sessions
         """
+        # Concurrent session creation can auto-start the same client. Keep the
+        # state check and all transport initialization under one lock so only
+        # one caller can spawn a runtime and install its connection at a time.
+        async with self._start_lock:
+            await self._start()
+
+    async def _start(self) -> None:
         if self._state == "connected":
             return
 

python/test_client_start.py

@@ -0,0 +1,85 @@
+"""Startup concurrency regressions without a live CLI runtime."""
+
+import asyncio
+from unittest.mock import AsyncMock
+
+import pytest
+
+from copilot import CopilotClient, RuntimeConnection
+
+
+@pytest.fixture
+def client():
+    client = CopilotClient(connection=RuntimeConnection.for_stdio(path="unused-cli"))
+    client._start_cli_server = AsyncMock()
+    client._connect_to_server = AsyncMock()
+    client._verify_protocol_version = AsyncMock()
+    return client
+
+
+@pytest.mark.parametrize(
+    "phase", ["_start_cli_server", "_connect_to_server", "_verify_protocol_version"]
+)
+async def test_concurrent_start_initializes_transport_once(client, phase):
+    entered = asyncio.Event()
+    release = asyncio.Event()
+
+    async def block():
+        entered.set()
+        await release.wait()
+
+    getattr(client, phase).side_effect = block
+    first = asyncio.create_task(client.start())
+    await asyncio.wait_for(entered.wait(), timeout=1)
+    second = asyncio.create_task(client.start())
+    try:
+        await asyncio.sleep(0)
+        assert not second.done()
+    finally:
+        release.set()
+        await asyncio.wait_for(asyncio.gather(first, second), timeout=1)
+
+    client._start_cli_server.assert_awaited_once()
+    client._connect_to_server.assert_awaited_once()
+    client._verify_protocol_version.assert_awaited_once()
+    await client.start()
+    client._start_cli_server.assert_awaited_once()
+
+
+async def test_start_can_retry_after_failure(client):
+    client._start_cli_server.side_effect = [RuntimeError("startup failed"), None]
+
+    with pytest.raises(RuntimeError, match="startup failed"):
+        await client.start()
+    await asyncio.wait_for(client.start(), timeout=1)
+
+    assert client._start_cli_server.await_count == 2
+    client._connect_to_server.assert_awaited_once()
+    client._verify_protocol_version.assert_awaited_once()
+
+
+async def test_cancelling_waiting_start_does_not_cancel_active_start(client):
+    entered = asyncio.Event()
+    release = asyncio.Event()
+
+    async def block():
+        entered.set()
+        await release.wait()
+
+    client._start_cli_server.side_effect = block
+    first = asyncio.create_task(client.start())
+    await asyncio.wait_for(entered.wait(), timeout=1)
+    second = asyncio.create_task(client.start())
+    try:
+        await asyncio.sleep(0)
+        second.cancel()
+        with pytest.raises(asyncio.CancelledError):
+            _ = await second
+        assert not first.done()
+    finally:
+        release.set()
+        await asyncio.wait_for(first, timeout=1)
+
+    await client.start()
+    client._start_cli_server.assert_awaited_once()
+    client._verify_protocol_version.assert_awaited_once()