Currently the ComputeManager abstract method create_compute_services only gives information from the alchemiscale server as to what tasks are available and how many compute services the server sees. However, to support compute services configured to claim many Tasks at once, it would be valuable for this method to include the max number of compute services this method can create to stay beneath the configured maximum.
We propose using claim_limit in ComputeServiceSettings in ComputeManager to determine how many services are allowed to be created in calls to create_compute_services so that downstream implementations can use this information in their scaling logic.
Currently the
ComputeManagerabstract methodcreate_compute_servicesonly gives information from thealchemiscaleserver as to what tasks are available and how many compute services the server sees. However, to support compute services configured to claim manyTasks at once, it would be valuable for this method to include the max number of compute services this method can create to stay beneath the configured maximum.We propose using
claim_limitinComputeServiceSettingsinComputeManagerto determine how many services are allowed to be created in calls tocreate_compute_servicesso that downstream implementations can use this information in their scaling logic.