tl;dr, I can get updates to your tools to you all more quickly.
Despite configuration files to the contrary, and significant amounts of documentation suggesting that LocalRunners are OK to use, during a personal conversation one of the developers was horrified to learn that people use it in production instances. Given that I have knowledge of HTCondor and its setup, I’ve switched us over to HTCondor. Previously, whenever I restarted galaxy, and jobs that were running would die. Restarts were necessary for tool updates. This resulted in a bad experience for users. We now run on a “local” cluster; jobs are dispatched to galaxy, I can restart galaxy as needed, and it’ll see that your job is done whenever it comes back.