[01:05:12] * bd808 off [22:23:47] You can make it a scheduled job triggering every minute, it will not overlap itself [22:41:45] isn't that just a hacky version of `--continuous`? [22:42:59] the specific (1) case from Brad is a job that should run all the time until is gets a "stop" signal from a different job [22:45:02] AnomieBOT implements a very robust job control and watchdog system that largely predates Toolforge, Grid Engine, Kubernetes, etc. It goes back to 2008 and Brad's closet. Over time he has adapted parts of it to grid and now k8s, but never ripped out his core system to depend instead on something outside [22:47:37] it supports deploy on push via graceful restarts, job status reconciliation via periodic watchdog checks, command and control message passing between jobs, and some other neat stuff [22:48:28] one of the goofy but fun things it does is self-publish it's code to wiki -- https://en.wikipedia.org/wiki/User:AnomieBOT/source [22:48:54] that's hilarious/scary [22:49:39] there is a git repo in the mix in the tool's $HOME [22:50:03] it uses the older bare repo pattern that we used to document on wikitech [22:55:26] * bd808 off to touch grass and walk a dog