To reduce the number of alerts caused by transient issues, I changed our monit config so we only get alerted if a default bridge fails three out of five consecutive checks.