Cron + a `pgrep` duplicate-run guard catches itself

Symptom

The job never runs. Not once. No error, no log. Cron records a clean exit 0.

Cause

The shell cron spawns has the script name on its command line, so pgrep hits every time → "already running" → immediate exit, every run.

Fix — PID file + kill -0

PIDFILE=/tmp/job.pid
if [ -f "$PIDFILE" ] && kill -0 "$(cat "$PIDFILE")" 2>/dev/null; then exit 0; fi
echo $$ > "$PIDFILE"
trap 'rm -f "$PIDFILE"' EXIT INT TERM HUP

kill -0 sends no signal; it only checks existence. A dead PID passes, so a stale file can't block you.

★ Check all three: 1) the file is non-empty (-s) 2) the content is numeric 3) that PID is alive

Later case — the silence read as "no orders"

The same guard sat on an order-notification loop. Cron logged a clean exit 0 every run, and the stretch with no notifications at all was read as "no notifications = no orders." A second, unrelated cause was blocking the same path that day, so fixing the guard brought nothing back.

★ When silence has two causes, fixing one changes nothing visible. If the silence survives your fix, suspect "there's another cause" before "the fix was wrong."

---

---