Based on our current setup (image bellow), even though we've set

num_ide_processes=1

, after some hours of inactivity the agent won't join the room, causing our frontend app to display an error. However, after a reload/retry, the agent is dispatched to the room. This behavior is reproducible both on a normal Linux server and in cloud‑agent deployments. How can we ensure the instances are always running and always able to spawn agents and dispatch them to rooms? @refined-appointment-81829 @tall-belgium-91876