fix: make the boot update actually fail loudly and survive long builds

The "update on reboot" unit silently did nothing useful. boot-update.sh piped
every command into tee, so `set -e` only saw tee's exit status: a failed
npm install or build left a half-built tree and still logged "Boot update
finished". It also used npm install (rewrites package-lock.json, so the next
git pull aborts), had no GIT_TERMINAL_PROMPT=0, and was not wrapped in a
function even though git pull rewrites it mid-run.

The unit's TimeoutStartSec=300 was the other half: npm ci + prisma generate +
next build takes far longer than 5 minutes in the container, so systemd
SIGTERMed the oneshot mid-install and left a node_modules tree npm could no
longer remove (ENOTEMPTY on rmdir node_modules/date-fns/isEqual).

Also resyncs package-lock.json with package.json, missed in the 1.115.0 commit.

v1.115.1

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

CHANGELOG.md

@@ -2,6 +2,28 @@
 
 All notable changes to this project are documented in this file.
 
+## [1.115.1] - 2026-09-03
+
+### Fixed
+- `scripts/boot-update.sh` (the "update on reboot" unit): every command was piped into `tee`,
+  so `set -e` only ever saw tee's exit status. A failed `npm install` or `npm run build` did
+  not stop the script and did not fail the systemd unit — it logged "Boot update finished"
+  and left a half-built tree. Now uses `set -euo pipefail`, logs through a single `exec`
+  redirect, and traps ERR.
+- `scripts/boot-update.sh`: switched from `npm install` to `npm ci`. Same fix update-app.sh
+  got in 1.114.1 — `npm install` rewrites package-lock.json, leaving the tree dirty so the
+  *next* `git pull` aborts.
+- `scripts/boot-update.sh`: body wrapped in a function so bash parses it in one go. The
+  script lives inside the checkout it pulls, so `git pull` rewriting it mid-run made bash
+  resume at a stale file offset.
+- `scripts/boot-update.sh`: set `GIT_TERMINAL_PROMPT=0` so a git credential prompt fails
+  fast instead of hanging the unit at boot, where there is no tty.
+- `scripts/my-weekly-todo-update.service`: raised `TimeoutStartSec` from 300 to 1800.
+  `npm ci` + `prisma generate` + `next build` takes far longer than 5 minutes in the
+  container; systemd SIGTERMed the oneshot unit mid-install, which left a `node_modules`
+  tree npm could no longer remove (`ENOTEMPTY: rmdir node_modules/date-fns/isEqual`).
+- `package-lock.json`: version resynced with `package.json` (missed in the 1.115.0 commit).
+
 ## [1.115.0] - 2026-09-03
 
 ### Added

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "my-weekly-todo-list",
-  "version": "1.114.2",
+  "version": "1.115.1",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "my-weekly-todo-list",
-      "version": "1.114.2",
+      "version": "1.115.1",
       "license": "MIT",
       "dependencies": {
         "@auth/prisma-adapter": "^2.11.1",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "my-weekly-todo-list",
-  "version": "1.115.0",
+  "version": "1.115.1",
   "description": "A web-based weekly task management application that organizes to-dos and calendar events in a single, intuitive weekly view",
   "main": "index.js",
   "scripts": {

scripts/boot-update.sh

@@ -1,66 +1,92 @@
 #!/bin/bash
-# Boot Update Script: Automatically pull latest main, build, and prepare for PM2
-# This script runs on LXC container boot BEFORE PM2 starts.
-# It ensures the app is always up-to-date with the main branch.
+#
+# Boot Update Script: pull latest main, install, build, and prepare for PM2.
+# Runs on LXC container boot BEFORE PM2 starts (my-weekly-todo-update.service).
+#
+# The whole body is wrapped in a function so bash parses it in one go: this script lives
+# inside the very checkout it updates, and `git pull` rewriting it mid-run would otherwise
+# make bash resume at a stale file offset.
 
-set -e
+main() {
+    # pipefail matters here: every command used to be piped into `tee`, so the shell saw
+    # tee's exit status (always 0). A failed npm install or build did not stop the script
+    # and did not fail the unit — it just logged "finished" and left a half-built tree.
+    set -euo pipefail
 
-APP_DIR="/root/My-Weekly-ToDo-List"
-LOG_FILE="/var/log/app-boot-update.log"
+    local APP_DIR="/root/My-Weekly-ToDo-List"
+    local LOG_FILE="/var/log/app-boot-update.log"
 
-log() {
-    echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1" | tee -a "$LOG_FILE"
-}
+    # Send everything to the log once, instead of piping each command through tee.
+    exec > >(tee -a "$LOG_FILE") 2>&1
+
+    log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1"; }
+
+    trap 'log "!!! Boot update FAILED at line $LINENO (exit $?) — not restarting the app"' ERR
+
+    # Fail fast instead of hanging on a credential prompt when run without a terminal
+    # (systemd has no tty). The host authenticates via a read-only SSH deploy key.
+    export GIT_TERMINAL_PROMPT=0
 
     log "=== Boot update started ==="
 
     cd "$APP_DIR"
 
-# Check and ensure we are on main branch
     log "Ensuring main branch..."
-git checkout main 2>&1 | tee -a "$LOG_FILE"
+    git checkout main
 
-# Check if there are remote changes
     log "Fetching from origin..."
-git fetch origin main 2>&1 | tee -a "$LOG_FILE"
+    git fetch origin main
 
+    local LOCAL REMOTE NEXT_DIR
     LOCAL=$(git rev-parse HEAD)
     REMOTE=$(git rev-parse origin/main)
     NEXT_DIR="$APP_DIR/.next"
 
     if [ "$LOCAL" = "$REMOTE" ] && [ -d "$NEXT_DIR" ]; then
-    log "Already up-to-date and .next directory exists. No rebuild needed."
+        log "Already up-to-date and .next exists. No rebuild needed."
         log "=== Boot update finished (no changes) ==="
-    exit 0
+        return 0
     fi
 
     if [ "$LOCAL" != "$REMOTE" ]; then
-    log "Changes detected. Updating from $LOCAL to "$REMOTE""
-    log "Pulling latest changes..."
-    git pull origin main 2>&1 | tee -a "$LOG_FILE"
+        log "Changes detected. Updating from $LOCAL to $REMOTE"
+        git pull origin main
 
+        # npm ci, not npm install: it installs exactly what package-lock.json specifies and
+        # never rewrites it. npm install rewrites the lockfile, which leaves the working
+        # tree dirty and makes the *next* git pull abort. npm ci also clears node_modules
+        # itself; if that ever fails with ENOTEMPTY, delete the directory by hand and rerun.
         log "Installing dependencies..."
-    npm install 2>&1 | tee -a "$LOG_FILE"
+        npm ci
     fi
 
     if [ ! -d "$NEXT_DIR" ] || [ "$LOCAL" != "$REMOTE" ]; then
-    log "Updating database schema (Safe Migrate)..."
-    npx prisma migrate deploy 2>&1 | tee -a "$LOG_FILE" || {
+        log "Updating database schema (safe migrate)..."
+        npx prisma migrate deploy || {
             log "migrate deploy failed, falling back to db push..."
-        npx prisma db push --accept-data-loss 2>&1 | tee -a "$LOG_FILE"
+            npx prisma db push --accept-data-loss
         }
 
         log "Regenerating Prisma client..."
-    npx prisma generate 2>&1 | tee -a "$LOG_FILE"
+        npx prisma generate
 
         log "Building application..."
-    npm run build 2>&1 | tee -a "$LOG_FILE"
+        npm run build
 
-    log "Preparing standalone directory (copying static assets and env)..."
-    cp -r public .next/standalone/ 2>&1 | tee -a "$LOG_FILE"
-    mkdir -p .next/standalone/.next/static 2>&1 | tee -a "$LOG_FILE"
-    cp -r .next/static .next/standalone/.next/ 2>&1 | tee -a "$LOG_FILE"
-    cp .env .next/standalone/ 2>/dev/null || true 2>&1 | tee -a "$LOG_FILE"
+        # next.config.js sets output: 'standalone', and Next.js does NOT copy public/ or
+        # .next/static/ into .next/standalone — that is a documented manual step. Skipping
+        # it ships new server code whose HTML references new chunk hashes while the old
+        # chunks sit in the standalone tree: they 404, React never hydrates, and every page
+        # hangs on "Loading session..." while the API still answers 200.
+        log "Preparing standalone directory (static assets and env)..."
+        cp -r public .next/standalone/
+        mkdir -p .next/standalone/.next/static
+        cp -r .next/static .next/standalone/.next/
+        cp .env .next/standalone/ 2>/dev/null || true
     fi
 
+    log "Deployed version: $(node -p "require('$APP_DIR/package.json').version")"
     log "=== Boot update finished (verified/rebuilt) ==="
+}
+
+main "$@"

scripts/my-weekly-todo-update.service

@@ -12,7 +12,10 @@ Environment=HOME=/root
 Environment=PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
 StandardOutput=journal+console
 StandardError=journal+console
-TimeoutStartSec=300
+# npm ci + prisma generate + next build in this container takes well over the old 300s.
+# systemd SIGTERMs a Type=oneshot unit at the timeout, which used to kill npm mid-install
+# and leave a node_modules tree that npm could no longer clean up (ENOTEMPTY on rmdir).
+TimeoutStartSec=1800
 
 [Install]
 WantedBy=multi-user.target