What is the impact if the SSM for an application is too busy to respond to a reclaim request?

In a large-scale cluster (5000 slots), the SSM is not able to release slots within the default EGO_DISCIPLINE_TIMEOUT period of 120 seconds when VEMKD reclaims a lot of slots. For large-scale clusters, it is recommended to set the EGO_DISCIPLINE_TIMEOUT parameter in the ego.conf file to 10 minutes. See ego.conf reference.