On the client side of a relayd-based wireless bridge using OpenWrt, I discovered there was a bug in that relayd version which made the process hang after it moved so many gigs of data. I made a cron job that pings the network relayd makes accessible. If the ping fails, it nukes relayd. Of course this relies on a live machine to ping. If this machine dies for some reason, the cron job would just keep killing relayd over and over again. 🥹