Disable the systemd timer for logrotate to ensure we don't conflict
This change disables and stops the logrotate timer in systemd for the haproxy nodes.
This appears to have been introduced in a recent version of systemd,
but causes us to have multiple instances of logrotate running at
midnight UTC on some nodes, especially where the minute to run the
crontab entry is 0
.
Ideally, we should eventually look to replace our crontab entries with the systemd unit/timers, and make the timer run hourly instead to match our current behaviour.