Update README.md

Fix notification-delivery cron docs: cron may not be installed, /var/log/ needs root

README.md

@@ -153,21 +153,38 @@ regardless of whether anything has actually delivered it yet.
 [stacks/infra/scripts/process-cron.sh](stacks/infra/scripts/process-cron.sh)
 calls both endpoints for **every** business under `STACKS_DIR` in one
 run, so a single crontab entry covers businesses added later too — no
-per-business cron line to maintain. Add this to the VPS's crontab
+per-business cron line to maintain.
-(`crontab -e`) once:
+
+Many minimal cloud VPS images (including some Debian ones) don't ship
+`cron` at all — `crontab -e` failing with `command not found` means this
+first:
+
+```bash
+sudo apt update && sudo apt install -y cron
+sudo systemctl enable --now cron
+```
+
+Then add this to the VPS's crontab (`crontab -e`) once — **not** `sudo
+crontab -e`: run it as the same non-root user that already owns
+`/opt/nextchair` (the one you've been running `git pull`/`docker build`
+as), so the log file below is writable without needing root:
 
 ```bash
 # Deliver queued notifications every 2 minutes.
-*/2 * * * * STACKS_DIR=/opt/nextchair/stacks /opt/nextchair/stacks/infra/scripts/process-cron.sh notifications >> /var/log/nextchair-cron.log 2>&1
+*/2 * * * * STACKS_DIR=/opt/nextchair/stacks /opt/nextchair/stacks/infra/scripts/process-cron.sh notifications >> /opt/nextchair/nextchair-cron.log 2>&1
 
 # Refresh Google Reviews once a day.
-0 3 * * * STACKS_DIR=/opt/nextchair/stacks /opt/nextchair/stacks/infra/scripts/process-cron.sh reviews >> /var/log/nextchair-cron.log 2>&1
+0 3 * * * STACKS_DIR=/opt/nextchair/stacks /opt/nextchair/stacks/infra/scripts/process-cron.sh reviews >> /opt/nextchair/nextchair-cron.log 2>&1
 ```
 
-(Adjust `STACKS_DIR` if you cloned this repo somewhere other than
+(Adjust `STACKS_DIR`/the log path if you cloned this repo somewhere other
-`/opt/nextchair`.) To confirm it's working, or to flush a notification
+than `/opt/nextchair`, or don't have write access to it — `/var/log/`
-that's stuck pending right now, run either line's command by hand and
+itself is **not** a safe default here: it needs root, and a crontab
-check `/var/log/nextchair-cron.log` — a healthy run logs `<slug>
+entry that can't write its own log file just silently produces no file
+at all, which looks identical to "the job never ran.") To confirm it's
+working, or to flush a notification that's stuck pending right now, run
+either line's command by hand and check
+`/opt/nextchair/nextchair-cron.log` — a healthy run logs `<slug>
 process-notifications ok` for every business.
 
 ## Updating after a code change