bot/src/bootstrap.js

@@ -4,11 +4,50 @@
 // container restart (crash, `docker compose restart`, host reboot) self-heals
 // without any admin-panel interaction. Node 20's built-in fetch is used; no
 // extra HTTP client dependency needed for a single startup call.
+//
+// The fetch RETRIES with backoff: on `docker compose up`, the bot and the app
+// start together and the bot's `depends_on: app` only waits for the container
+// to *start*, not for the app's internal server to be listening (it still has
+// to reach the DB and boot Express). Without retries the very first fetch loses
+// that race, bootstrap gives up, and the bot sits disconnected while the DB
+// still says enabled — the exact "enabled but disconnected until I toggle it"
+// bug. Retrying until the site answers makes a cold whole-stack start heal on
+// its own.
 const discordManager = require('./discord/discordManager')
 const createLogger = require('./utils/logger')
 
 const log = createLogger('bootstrap')
 
+const MAX_ATTEMPTS = 30 // ~30 tries * ~2s ≈ 1 min of patience for the app to come up
+const RETRY_DELAY_MS = 2000
+
+const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms))
+
+// Fetch config from the site, retrying while the site is unreachable or not yet
+// ready (network error or 5xx). Returns the parsed config, or null if we gave
+// up after MAX_ATTEMPTS. A 4xx (e.g. bad internal key) is a real misconfig, not
+// a transient startup race, so we don't retry those.
+async function fetchConfig(siteUrl, key) {
+  for (let attempt = 1; attempt <= MAX_ATTEMPTS; attempt += 1) {
+    try {
+      const res = await fetch(siteUrl, { headers: { 'X-Internal-Key': key } })
+      if (res.ok) return await res.json()
+      if (res.status >= 400 && res.status < 500) {
+        log.error('boot-time config fetch rejected — not retrying', { status: res.status })
+        return null
+      }
+      log.warn('boot-time config fetch not ready — retrying', { status: res.status, attempt })
+    } catch (err) {
+      log.warn('boot-time config fetch errored — retrying', { message: err.message, attempt })
+    }
+    if (attempt < MAX_ATTEMPTS) await sleep(RETRY_DELAY_MS)
+  }
+  log.error('boot-time config fetch gave up after retries — staying disconnected until the admin panel pushes config', {
+    attempts: MAX_ATTEMPTS,
+  })
+  return null
+}
+
 async function bootstrap() {
   const siteUrl = process.env.SITE_INTERNAL_URL
   const key = process.env.BOT_INTERNAL_KEY
@@ -17,21 +56,18 @@ async function bootstrap() {
     return
   }
 
-  try {
-    const res = await fetch(siteUrl, { headers: { 'X-Internal-Key': key } })
-    if (!res.ok) {
-      log.error('boot-time config fetch failed', { status: res.status })
-      return
-    }
-    const config = await res.json()
-    if (config.enabled) {
-      log.info('boot-time config says enabled — reconnecting', { guildId: config.guildId })
+  const config = await fetchConfig(siteUrl, key)
+  if (!config) return
+
+  if (config.enabled) {
+    log.info('boot-time config says enabled — reconnecting', { guildId: config.guildId })
+    try {
       await discordManager.start({ token: config.token, guildId: config.guildId })
-    } else {
-      log.info('boot-time config says disabled — staying disconnected')
+    } catch (err) {
+      log.error('boot-time reconnect failed', { message: err.message })
     }
-  } catch (err) {
-    log.error('boot-time config fetch errored', { message: err.message })
+  } else {
+    log.info('boot-time config says disabled — staying disconnected')
   }
 }