opensource/n8n
3
0
Fork 0
mirror of https://github.com/n8n-io/n8n.git synced 2026-10-13 09:14:41 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: Local pnpm nathan command usage (no-changelog) (#39385)

Browse Source
This commit is contained in:
Matsu
2026-09-23 12:25:21 +00:00
committed by GitHub
parent 042d16718e
commit 3b4da4f39b
2 changed files with 63 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.agents/skills/nathan/SKILL.md
+5 -5
View File
@@ -13,11 +13,11 @@ the same commands from the repo via `pnpm nathan`.
## Prerequisites
- **No tunnel setup needed.** Nathan replies asynchronously through a short-lived
public tunnel that the script opens for you (`npx localtunnel`). A `deploy` to a
**new** `test-<name>` also polls that instance URL directly, so it still reports
success even if the tunnel drops. (Redeploying a name that's *already up* skips
the poll — it can't tell the old instance from the new — so it falls back to the
tunnel; prefer a fresh name when you need the reliable report.)
Cloudflare Quick Tunnel that the script opens for you (`npx cloudflared`). A
`deploy` to a **new** `test-<name>` also polls that instance URL directly, so it
still reports success even if the tunnel drops. (Redeploying a name that's
*already up* skips the poll — it can't tell the old instance from the new — so
it falls back to the tunnel; prefer a fresh name when you need the reliable report.)
- **A token in `~/.n8n/dev/nathan-token`.** If a command reports no token, **ask the
user for one** — point them at the form
(`https://internal.users.n8n.cloud/form/d6d34a2f-4899-4ee8-afc8-f8c41a8a243d`),
scripts/nathan.mjs
+58 -14
View File
@@ -14,7 +14,7 @@
// Note: --license enterprise and AI (--ai) features cannot be used together.
//
// Needs a token in ~/.n8n/dev/nathan-token — on first run it links you to a form to
// get one and saves it there. A public tunnel is opened via `npx localtunnel`.
// get one and saves it there. A public tunnel is opened via `npx cloudflared`.
import http from 'node:http';
import fs from 'node:fs';
import os from 'node:os';
@@ -38,6 +38,7 @@ const IDLE_MS = posNum(process.env.NATHAN_IDLE_MS, 8000, 'NATHAN_IDLE_MS'); // g
// 15-20 min, so keep this generous; override via NATHAN_TIMEOUT_MS.
const TIMEOUT_MS = posNum(process.env.NATHAN_TIMEOUT_MS, 1_500_000, 'NATHAN_TIMEOUT_MS'); // 25 min
const TUNNEL_START_MS = 45000;
const TUNNEL_DNS_SETTLE_MS = 10000;
const STAGING_DOMAIN = 'stage-app.n8n.cloud'; // instances live at https://<name>.<domain>
const DEFAULT_SLACK_CHANNEL = 'C0BGVHZ0SCW'; // #updates-pnpm-nathan
const DEFAULT_SLACK_CHANNEL_URL = 'https://n8nio.slack.com/archives/C0BGVHZ0SCW';
@@ -157,7 +158,15 @@ const CALLBACK_PATH = `/cb/${randomUUID()}`;
const MAX_BODY = 1_000_000; // Nathan's replies are small Slack messages
let done, doneReason, settled = false;
const finished = new Promise((r) => (done = r));
const finish = (reason) => { if (settled) return; settled = true; doneReason = reason; done(); };
const finish = (reason) => {
if (settled) {
return;
}
settled = true;
doneReason = reason;
done();
};
let idleTimer;
const server = http.createServer((req, res) => {
// A GET is our own reachability probe (or noise) — ack it, don't process it.
@@ -167,7 +176,7 @@ const server = http.createServer((req, res) => {
let raw = '', tooBig = false;
req.on('data', (c) => {
if (tooBig) return;
raw += c;
raw += c;https://github.com/n8n-io/n8n-argo-apps/pull/644
if (raw.length > MAX_BODY) { tooBig = true; res.writeHead(413).end('payload too large'); req.destroy(); }
});
req.on('end', () => {
@@ -185,29 +194,55 @@ const server = http.createServer((req, res) => {
return; // otherwise work is starting; wait (overall timeout backstops)
}
// A known error reply must fail the run; anything else is a success.
idleTimer = setTimeout(() => finish(isError(body) ? 'error' : 'idle'), IDLE_MS);
idleTimer = setTimeout(() => {
console.log("Idle timer tripped")
finish(isError(body) ? 'error' : 'idle')
}, IDLE_MS);
});
});
await new Promise((r) => server.listen(0, '127.0.0.1', r));
const port = server.address().port;
// --- public tunnel via npx localtunnel ---------------------------------------
// Pin an exact, vetted version — an unpinned `npx localtunnel` would run whatever
// --- public tunnel via npx cloudflared ---------------------------------------
// Pin exact, vetted versions — an unpinned `npx cloudflared` would run whatever
// the registry currently serves, as the user, with read access to the token file.
// To update: bump this version deliberately after reviewing the release.
const LOCALTUNNEL_VERSION = '2.0.2';
// To update: bump these versions deliberately after reviewing both releases.
const CLOUDFLARED_NPM_VERSION = '0.7.3';
const CLOUDFLARED_VERSION = '2026.9.1';
function startTunnel() {
const proc = spawn('npx', ['-y', `localtunnel@${LOCALTUNNEL_VERSION}`, '--port', String(port)], { stdio: ['ignore', 'pipe', 'pipe'] });
const urlRe = /https:\/\/[^\s]+\.loca\.lt/;
const proc = spawn(
'npx',
[
'-y',
`cloudflared@${CLOUDFLARED_NPM_VERSION}`,
'tunnel',
'--url',
`http://127.0.0.1:${port}`,
'--no-autoupdate',
],
{
stdio: ['ignore', 'pipe', 'pipe'],
env: { ...process.env, CLOUDFLARED_VERSION },
},
);
const urlRe = /https:\/\/[^\s]+\.trycloudflare\.com/;
return new Promise((resolve, reject) => {
const t = setTimeout(() => reject(new Error('tunnel did not start in time')), TUNNEL_START_MS);
let url;
let connected = false;
let output = '';
const scan = (buf) => {
const m = String(buf).match(urlRe);
if (m) { clearTimeout(t); resolve({ url: m[0], proc }); }
output = (output + String(buf)).slice(-10000);
url ??= output.match(urlRe)?.[0];
connected ||= output.includes('Registered tunnel connection');
if (url && connected) {
clearTimeout(t);
resolve({ url, proc });
}
};
proc.stdout.on('data', scan);
proc.stderr.on('data', scan);
proc.on('exit', (code) => reject(new Error(`localtunnel process exited (${code})`)));
proc.on('exit', (code) => reject(new Error(`cloudflared process exited (${code})`)));
});
}
@@ -215,6 +250,8 @@ function startTunnel() {
// so firing immediately can make Nathan's callback fail with ENOTFOUND. Self-probe
// the tunnel (GET, ignored by the server above) until it routes back to us.
async function waitReachable(url) {
// Avoid caching NXDOMAIN before Cloudflare publishes the Quick Tunnel hostname.
await new Promise((r) => setTimeout(r, TUNNEL_DNS_SETTLE_MS));
const deadline = Date.now() + TUNNEL_START_MS;
while (Date.now() < deadline) {
try { if ((await fetch(url, { signal: AbortSignal.timeout(5000) })).ok) return true; }
@@ -263,6 +300,7 @@ if (!res.ok) {
console.error(`Webhook returned HTTP ${res.status}`);
await cleanup(1);
}
console.error('Sent. Waiting for Nathan to reply (Ctrl-C to stop)…\n');
// Reliable backstop: if the instance URL is predictable, poll it directly so a
@@ -277,7 +315,9 @@ if (pollUrl) {
(async () => {
while (!settled) {
if (await up(pollUrl)) {
if (!settled) console.log(`\n${'─'.repeat(60)}\n✅ Instance is up: ${pollUrl}\n Login: test@n8n.io / helloWorld7 (default test owner)\n`);
if (!settled) {
console.log(`\n${'─'.repeat(60)}\n✅ Instance is up: ${pollUrl}\n Waiting for workflow to report login credentials...\n`);
}
return finish('up');
}
await new Promise((r) => setTimeout(r, 10000));
@@ -291,12 +331,16 @@ process.on('SIGINT', () => finish('interrupted'));
await finished;
clearTimeout(overall);
// Wait for Nathan workflow to finish sending messages to us
await new Promise(r => setTimeout(r, 15000))
if (doneReason === 'timeout') console.error('\n⏱ Timed out waiting for the final reply — the deploy may still be running (check Slack / Grafana).');
if (doneReason === 'interrupted') console.error('\nStopped. The command may still be running on Nathan.');
if (doneReason === 'error') console.error('\n✖ Nathan reported an error (see the reply above).');
await cleanup(['idle', 'up', 'local'].includes(doneReason) ? 0 : 1);
function cleanup(code) {
console.log("Closing server and tunnel.");
try { tunnel?.proc.kill(); } catch {}
server.close();
// give the tunnel a beat to die, then exit hard (server/proc keep the loop alive)