fix(mosaic): resumable gateway install + prominent admin token (#393)

docs/scratchpads/gateway-install-ux-20260404.md

@@ -0,0 +1,47 @@
+# Gateway Install UX Fixes — 2026-04-04
+
+## Context
+
+User hit two publish-drift bugs on `mosaic gateway install` (fixed in #386 and
+#389). On top of those, the install UX itself has three concrete problems:
+
+1. Admin API token is generated by the bootstrap step and saved to
+   `~/.config/mosaic/gateway/meta.json`, but never shown to the user. There is
+   no way to retrieve it without reading the file or running `mosaic gateway
+config` afterward.
+2. When install crashes mid-way (e.g. daemon fails to become healthy), the
+   next run of `mosaic gateway install` prompts "Reinstall? [y/N]". Saying N
+   aborts, leaving the user with a half-configured install and no clear path
+   forward. There is no resume path.
+3. Health-check failure only prints "Check logs: mosaic gateway logs" — forcing
+   the user to run another command to see the actual error.
+
+## Goal
+
+Make `mosaic gateway install` a single end-to-end, resumable command that shows
+the admin token on creation and surfaces errors inline.
+
+## Plan
+
+- `install.ts`: detect partial state (meta exists, no admin token, daemon not
+  running) and skip already-completed phases instead of aborting.
+- `install.ts`: add a prominent "Admin API Token" banner printed right after
+  bootstrap succeeds. Include copy-now warning.
+- `install.ts`: on health-check timeout, read the tail of `LOG_FILE` and print
+  the last ~30 non-empty lines before returning.
+- Keep `mosaic gateway config` as-is (view/edit env vars); this is not the
+  setup wizard.
+- No new commands. No new flags yet.
+
+## Out of scope
+
+- Readline/stdin piping fragility (pre-existing, not related to user complaint).
+- Refactor to a phased state machine (overkill for three targeted fixes).
+
+## Verification
+
+- Typecheck, lint, format (mandatory gates).
+- Manual end-to-end: fresh install → confirm token displayed.
+- Manual resume: delete daemon.pid mid-install → re-run → confirm it resumes.
+- Manual failure: point ENV_FILE port to a used port → re-run → confirm log
+  tail is printed.

packages/mosaic/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mosaicstack/mosaic",
-  "version": "0.0.19",
+  "version": "0.0.20",
   "repository": {
     "type": "git",
     "url": "https://git.mosaicstack.dev/mosaic/mosaic-stack.git",

packages/mosaic/src/commands/gateway/install.ts

@@ -1,21 +1,26 @@
 import { randomBytes } from 'node:crypto';
-import { writeFileSync } from 'node:fs';
+import { existsSync, readFileSync, writeFileSync } from 'node:fs';
 import { join } from 'node:path';
 import { createInterface } from 'node:readline';
 import type { GatewayMeta } from './daemon.js';
 import {
   ENV_FILE,
   GATEWAY_HOME,
+  LOG_FILE,
   ensureDirs,
+  getDaemonPid,
   installGatewayPackage,
   readMeta,
   resolveGatewayEntry,
   startDaemon,
+  stopDaemon,
   waitForHealth,
   writeMeta,
   getInstalledGatewayVersion,
 } from './daemon.js';
 
+const MOSAIC_CONFIG_FILE = join(GATEWAY_HOME, 'mosaic.config.json');
+
 interface InstallOpts {
   host: string;
   port: number;
@@ -36,30 +41,198 @@ export async function runInstall(opts: InstallOpts): Promise<void> {
 }
 
 async function doInstall(rl: ReturnType<typeof createInterface>, opts: InstallOpts): Promise<void> {
-  // Check existing installation
   const existing = readMeta();
-  if (existing) {
-    const answer = await prompt(
-      rl,
-      `Gateway already installed (v${existing.version}). Reinstall? [y/N] `,
+  const envExists = existsSync(ENV_FILE);
+  const mosaicConfigExists = existsSync(MOSAIC_CONFIG_FILE);
+  let hasConfig = envExists && mosaicConfigExists;
+  let daemonRunning = getDaemonPid() !== null;
+  const hasAdminToken = Boolean(existing?.adminToken);
+  // `opts.host` already incorporates meta fallback via the parent command
+  // in gateway.ts (resolveOpts). Using it directly also lets a user pass
+  // `--host X` to recover from a previous install that stored a broken
+  // host. We intentionally do not prefer `existing.host` over `opts.host`.
+  const host = opts.host;
+
+  // Corrupt partial state: exactly one of the two config files survived.
+  // This happens when an earlier install was interrupted between writing
+  // .env and mosaic.config.json. Rewriting the missing one would silently
+  // rotate BETTER_AUTH_SECRET or clobber saved DB/Valkey URLs. Refuse to
+  // guess — tell the user how to recover. Check file presence only; do
+  // NOT gate on `existing`, because the installer writes config before
+  // meta, so an interrupted first install has no meta yet.
+  if ((envExists || mosaicConfigExists) && !hasConfig) {
+    console.error('Gateway install is in a corrupt partial state:');
+    console.error(`  .env file:            ${envExists ? 'present' : 'MISSING'} (${ENV_FILE})`);
+    console.error(
+      `  mosaic.config.json:   ${mosaicConfigExists ? 'present' : 'MISSING'} (${MOSAIC_CONFIG_FILE})`,
     );
-    if (answer.toLowerCase() !== 'y') {
-      console.log('Aborted.');
-      return;
-    }
+    console.error('\nRun `mosaic gateway uninstall` to clean up, then re-run install.');
+    return;
   }
 
-  // Step 1: Install npm package
-  if (!opts.skipInstall) {
+  // Fully set up already — offer to re-run the config wizard and restart.
+  // The wizard allows changing storage tier / DB URLs, so this can move
+  // the install onto a different data store. We do NOT wipe persisted
+  // local data here — for a true scratch wipe run `mosaic gateway
+  // uninstall` first.
+  let explicitReinstall = false;
+  if (existing && hasConfig && daemonRunning && hasAdminToken) {
+    console.log(`Gateway is already installed and running (v${existing.version}).`);
+    console.log(`  Endpoint:  http://${existing.host}:${existing.port.toString()}`);
+    console.log(`  Status:    mosaic gateway status`);
+    console.log();
+    console.log('Re-running the config wizard will:');
+    console.log('  - regenerate .env and mosaic.config.json');
+    console.log('  - restart the daemon');
+    console.log('  - preserve BETTER_AUTH_SECRET (sessions stay valid)');
+    console.log('  - clear the stored admin token (you will re-bootstrap an admin user)');
+    console.log('  - allow changing storage tier / DB URLs (may point at a different data store)');
+    console.log('To wipe persisted data, run `mosaic gateway uninstall` first.');
+    const answer = await prompt(rl, 'Re-run config wizard? [y/N] ');
+    if (answer.trim().toLowerCase() !== 'y') {
+      console.log('Nothing to do.');
+      return;
+    }
+    // Fall through. The daemon stop below triggers because hasConfig=false
+    // forces the wizard to re-run.
+    hasConfig = false;
+    explicitReinstall = true;
+  } else if (existing && (hasConfig || daemonRunning)) {
+    // Partial install detected — resume instead of re-prompting the user.
+    console.log('Detected a partial gateway installation — resuming setup.\n');
+  }
+
+  // If we are going to (re)write config, the running daemon would end up
+  // serving the old config while health checks and meta point at the new
+  // one. Always stop the daemon before writing config.
+  if (!hasConfig && daemonRunning) {
+    console.log('Stopping gateway daemon before writing new config...');
+    try {
+      await stopDaemon();
+    } catch (err) {
+      const msg = err instanceof Error ? err.message : String(err);
+      if (/not running/i.test(msg)) {
+        // Raced with daemon exit — fine, proceed.
+      } else {
+        console.error(`Failed to stop running daemon: ${msg}`);
+        console.error('Refusing to rewrite config while an unknown-state daemon is running.');
+        console.error('Stop it manually (mosaic gateway stop) and re-run install.');
+        return;
+      }
+    }
+    // Re-check — stop may have succeeded but we want to be sure before
+    // writing new config files and starting a fresh process.
+    if (getDaemonPid() !== null) {
+      console.error('Gateway daemon is still running after stop attempt. Aborting.');
+      return;
+    }
+    daemonRunning = false;
+  }
+
+  // Step 1: Install npm package. Always run on first install and on any
+  // resume where the daemon is NOT already running — a prior failure may
+  // have been caused by a broken package version, and the retry should
+  // pick up the latest release. Skip only when resuming while the daemon
+  // is already alive (package must be working to have started).
+  if (!opts.skipInstall && !daemonRunning) {
     installGatewayPackage();
   }
 
   ensureDirs();
 
-  // Step 2: Collect configuration
+  // Step 2: Collect configuration (skip if both files already exist).
+  // On resume, treat the .env file as authoritative for port — but let a
+  // user-supplied non-default `--port` override it so they can recover
+  // from a conflicting saved port the same way `--host` lets them
+  // recover from a bad saved host. `opts.port === 14242` is commander's
+  // default (not explicit user input), so we prefer .env in that case.
+  let port: number;
+  const regeneratedConfig = !hasConfig;
+  if (hasConfig) {
+    const envPort = readPortFromEnv();
+    port = opts.port !== 14242 ? opts.port : (envPort ?? existing?.port ?? opts.port);
+    console.log(`Using existing config at ${ENV_FILE} (port ${port.toString()})`);
+  } else {
+    port = await runConfigWizard(rl, opts);
+  }
+
+  // Step 3: Write meta.json. Prefer host from existing meta when resuming.
+  let entryPoint: string;
+  try {
+    entryPoint = resolveGatewayEntry();
+  } catch {
+    console.error('Error: Gateway package not found after install.');
+    console.error('Check that @mosaicstack/gateway installed correctly.');
+    return;
+  }
+
+  const version = getInstalledGatewayVersion() ?? 'unknown';
+  // Preserve the admin token only on a pure resume (no config regeneration).
+  // Any time we regenerated config, the wizard may have pointed at a
+  // different storage tier / DB URL, so the old token is unverifiable —
+  // drop it and require re-bootstrap.
+  const preserveToken = !regeneratedConfig && Boolean(existing?.adminToken);
+  const meta: GatewayMeta = {
+    version,
+    installedAt: explicitReinstall
+      ? new Date().toISOString()
+      : (existing?.installedAt ?? new Date().toISOString()),
+    entryPoint,
+    host,
+    port,
+    ...(preserveToken && existing?.adminToken ? { adminToken: existing.adminToken } : {}),
+  };
+  writeMeta(meta);
+
+  // Step 4: Start the daemon (idempotent — skip if already running).
+  if (!daemonRunning) {
+    console.log('\nStarting gateway daemon...');
+    try {
+      const pid = startDaemon();
+      console.log(`Gateway started (PID ${pid.toString()})`);
+    } catch (err) {
+      console.error(`Failed to start: ${err instanceof Error ? err.message : String(err)}`);
+      printLogTail();
+      return;
+    }
+  } else {
+    console.log('\nGateway daemon is already running.');
+  }
+
+  // Step 5: Wait for health
+  console.log('Waiting for gateway to become healthy...');
+  const healthy = await waitForHealth(host, port, 30_000);
+  if (!healthy) {
+    console.error('\nGateway did not become healthy within 30 seconds.');
+    printLogTail();
+    console.error('\nFix the underlying error above, then re-run `mosaic gateway install`.');
+    return;
+  }
+  console.log('Gateway is healthy.\n');
+
+  // Step 6: Bootstrap — first admin user.
+  await bootstrapFirstUser(rl, host, port, meta);
+
+  console.log('\n─── Installation Complete ───');
+  console.log(`  Endpoint:  http://${host}:${port.toString()}`);
+  console.log(`  Config:    ${GATEWAY_HOME}`);
+  console.log(`  Logs:      mosaic gateway logs`);
+  console.log(`  Status:    mosaic gateway status`);
+}
+
+async function runConfigWizard(
+  rl: ReturnType<typeof createInterface>,
+  opts: InstallOpts,
+): Promise<number> {
   console.log('\n─── Gateway Configuration ───\n');
 
-  // Tier selection
+  // If a previous .env exists on disk, reuse its BETTER_AUTH_SECRET so
+  // regenerating config does not silently log out existing users.
+  const preservedAuthSecret = readEnvVarFromFile('BETTER_AUTH_SECRET');
+  if (preservedAuthSecret) {
+    console.log('(Preserving existing BETTER_AUTH_SECRET — auth sessions will remain valid.)\n');
+  }
+
   console.log('Storage tier:');
   console.log('  1. Local (embedded database, no dependencies)');
   console.log('  2. Team (PostgreSQL + Valkey required)');
@@ -91,10 +264,8 @@ async function doInstall(rl: ReturnType<typeof createInterface>, opts: InstallOp
   const corsOrigin =
     (await prompt(rl, 'CORS origin [http://localhost:3000]: ')) || 'http://localhost:3000';
 
-  // Generate auth secret
-  const authSecret = randomBytes(32).toString('hex');
+  const authSecret = preservedAuthSecret ?? randomBytes(32).toString('hex');
 
-  // Step 3: Write .env
   const envLines = [
     `GATEWAY_PORT=${port.toString()}`,
     `BETTER_AUTH_SECRET=${authSecret}`,
@@ -116,7 +287,6 @@ async function doInstall(rl: ReturnType<typeof createInterface>, opts: InstallOp
   writeFileSync(ENV_FILE, envLines.join('\n') + '\n', { mode: 0o600 });
   console.log(`\nConfig written to ${ENV_FILE}`);
 
-  // Step 3b: Write mosaic.config.json
   const mosaicConfig =
     tier === 'local'
       ? {
@@ -132,66 +302,81 @@ async function doInstall(rl: ReturnType<typeof createInterface>, opts: InstallOp
           memory: { type: 'pgvector' },
         };
 
-  const configFile = join(GATEWAY_HOME, 'mosaic.config.json');
-  writeFileSync(configFile, JSON.stringify(mosaicConfig, null, 2) + '\n', { mode: 0o600 });
-  console.log(`Config written to ${configFile}`);
+  writeFileSync(MOSAIC_CONFIG_FILE, JSON.stringify(mosaicConfig, null, 2) + '\n', { mode: 0o600 });
+  console.log(`Config written to ${MOSAIC_CONFIG_FILE}`);
 
-  // Step 4: Write meta.json
-  let entryPoint: string;
+  return port;
+}
+
+function readEnvVarFromFile(key: string): string | null {
+  if (!existsSync(ENV_FILE)) return null;
   try {
-    entryPoint = resolveGatewayEntry();
+    for (const line of readFileSync(ENV_FILE, 'utf-8').split('\n')) {
+      const trimmed = line.trim();
+      if (!trimmed || trimmed.startsWith('#')) continue;
+      const eqIdx = trimmed.indexOf('=');
+      if (eqIdx <= 0) continue;
+      if (trimmed.slice(0, eqIdx) !== key) continue;
+      return trimmed.slice(eqIdx + 1);
+    }
   } catch {
-    console.error('Error: Gateway package not found after install.');
-    console.error('Check that @mosaicstack/gateway installed correctly.');
+    return null;
+  }
+  return null;
+}
+
+function readPortFromEnv(): number | null {
+  const raw = readEnvVarFromFile('GATEWAY_PORT');
+  if (raw === null) return null;
+  const parsed = parseInt(raw, 10);
+  return Number.isNaN(parsed) ? null : parsed;
+}
+
+function printLogTail(maxLines = 30): void {
+  if (!existsSync(LOG_FILE)) {
+    console.error(`(no log file at ${LOG_FILE})`);
     return;
   }
-
-  const version = getInstalledGatewayVersion() ?? 'unknown';
-
-  const meta = {
-    version,
-    installedAt: new Date().toISOString(),
-    entryPoint,
-    host: opts.host,
-    port,
-  };
-  writeMeta(meta);
-
-  // Step 5: Start the daemon
-  console.log('\nStarting gateway daemon...');
   try {
-    const pid = startDaemon();
-    console.log(`Gateway started (PID ${pid.toString()})`);
+    const lines = readFileSync(LOG_FILE, 'utf-8')
+      .split('\n')
+      .filter((l) => l.trim().length > 0);
+    const tail = lines.slice(-maxLines);
+    if (tail.length === 0) {
+      console.error('(log file is empty)');
+      return;
+    }
+    console.error(`\n─── Last ${tail.length.toString()} log lines (${LOG_FILE}) ───`);
+    for (const line of tail) console.error(line);
+    console.error('─────────────────────────────────────────────');
   } catch (err) {
-    console.error(`Failed to start: ${err instanceof Error ? err.message : String(err)}`);
-    return;
+    console.error(`Could not read log file: ${err instanceof Error ? err.message : String(err)}`);
   }
+}
 
-  // Step 6: Wait for health
-  console.log('Waiting for gateway to become healthy...');
-  const healthy = await waitForHealth(opts.host, port, 30_000);
-  if (!healthy) {
-    console.error('Gateway did not become healthy within 30 seconds.');
-    console.error(`Check logs: mosaic gateway logs`);
-    return;
-  }
-  console.log('Gateway is healthy.\n');
-
-  // Step 7: Bootstrap — first user setup
-  await bootstrapFirstUser(rl, opts.host, port, meta);
-
-  console.log('\n─── Installation Complete ───');
-  console.log(`  Endpoint:  http://${opts.host}:${port.toString()}`);
-  console.log(`  Config:    ${GATEWAY_HOME}`);
-  console.log(`  Logs:      mosaic gateway logs`);
-  console.log(`  Status:    mosaic gateway status`);
+function printAdminTokenBanner(token: string): void {
+  const border = '═'.repeat(68);
+  console.log();
+  console.log(border);
+  console.log('  Admin API Token');
+  console.log(border);
+  console.log();
+  console.log(`    ${token}`);
+  console.log();
+  console.log('  Save this token now — it will not be shown again in full.');
+  console.log('  It is stored (read-only) at:');
+  console.log(`    ${join(GATEWAY_HOME, 'meta.json')}`);
+  console.log();
+  console.log('  Use it with admin endpoints, e.g.:');
+  console.log(`    mosaic gateway --token <token> status`);
+  console.log(border);
 }
 
 async function bootstrapFirstUser(
   rl: ReturnType<typeof createInterface>,
   host: string,
   port: number,
-  meta: Omit<GatewayMeta, 'adminToken'> & { adminToken?: string },
+  meta: GatewayMeta,
 ): Promise<void> {
   const baseUrl = `http://${host}:${port.toString()}`;
 
@@ -201,7 +386,12 @@ async function bootstrapFirstUser(
 
     const status = (await statusRes.json()) as { needsSetup: boolean };
     if (!status.needsSetup) {
-      console.log('Admin user already exists — skipping setup.');
+      if (meta.adminToken) {
+        console.log('Admin user already exists (token on file).');
+      } else {
+        console.log('Admin user already exists — skipping setup.');
+        console.log('(No admin token on file — sign in via the web UI to manage tokens.)');
+      }
       return;
     }
   } catch {
@@ -247,12 +437,12 @@ async function bootstrapFirstUser(
       token: { plaintext: string };
     };
 
-    // Save admin token to meta
+    // Persist the token so future CLI calls can authenticate automatically.
     meta.adminToken = result.token.plaintext;
-    writeMeta(meta as GatewayMeta);
+    writeMeta(meta);
 
     console.log(`\nAdmin user created: ${result.user.email}`);
-    console.log('Admin API token saved to gateway config.');
+    printAdminTokenBanner(result.token.plaintext);
   } catch (err) {
     console.error(`Bootstrap error: ${err instanceof Error ? err.message : String(err)}`);
   }