If I'm not mistaken, you can disable write-in to execution history from both launch services and scheduled jobs (because of exactly this reason - frequently running jobs).
But in your case I would not bother and just check server's HTTP API, more specifically cluster_status
; if that one will get unresponsive, your server is down. See documentation
for more info. Disregard its name, it works for single node too.