Reminds me of what patio11 does (off the top of my head, he rolled his own solution using Twilio). Yep. I can't recommend this, by the way. (+) Use PagerDut

Reminds me of what patio11 does (off the top of my head, he rolled his own solution using Twilio). Yep. I can't recommend this, by the way. (+) Use PagerDuty or something. + You run the risk of losing the monitoring infrastructure and the rest of your infrastructure to the same cause, thus not knowing that the system is down. If you implement it yourself, do a bidirectional deadman's switch, where the website raises a yellow alert if example.net/monitoring-endpoint isn't hit in e.g. 15 minutes and the monitoring infrastructure raises a red alert if the monitoring endpoint doesn't respond with a green result. (Mine runs a bunch of tests that demonstrate minimum functioning of the system and verifies that user activity within the last interval is within historical norms +/- a fudge factor. If anything goes wrong, it 500s that page.)