Been running it in production on aws and gcp for years now (self hosted). Imo whatever seconds you’ll shave off by skipping scheduler loop will be dwarfed by no

Been running it in production on aws and gcp for years now (self hosted). Imo whatever seconds you’ll shave off by skipping scheduler loop will be dwarfed by nodes not being available / slow to come which happens often and for which there’s no sla