Running PHP scripts from cron: paths, output and common mistakes

Scripts that work in a browser often fail under cron, because the environment is different.

1. Use the full PHP path

/usr/local/bin/php /home/USERNAME/domains/yourdomain.co.uk/public_html/tasks/run.php

To confirm the PHP path and version cron uses, create a one-off job:

which php > /home/USERNAME/cron-test.txt; php -v >> /home/USERNAME/cron-test.txt

Read the file in the File Manager, then delete the job.

2. Don’t rely on the working directory

Cron doesn’t run from the script’s folder. Build paths from the script’s own location:

<?php
require __DIR__ . '/../config/bootstrap.php';

3. No web variables

$_SERVER['HTTP_HOST'], $_GET and sessions don’t exist. Pass settings as arguments instead:

/usr/local/bin/php /path/to/run.php --task=cleanup

4. Stop overlapping runs

If a job can take longer than its interval, prevent two copies running at once:

<?php
$lock = fopen(__DIR__ . '/run.lock', 'c');
if (!flock($lock, LOCK_EX | LOCK_NB)) {
    exit(0); // previous run still going
}
// ... work ...

5. Log what happens

/usr/local/bin/php /path/to/run.php >> /home/USERNAME/logs/run.log 2>&1

Include timestamps in your output so the log is useful.

6. Keep cron scripts out of public reach

Put them outside public_html, or block web access, so a visitor can’t trigger them by URL.

Alternative: calling a URL

If a job must run through the web server:

curl -fsS https://yourdomain.co.uk/cron.php?key=LONG_SECRET >/dev/null

Protect the URL with a secret key.