Fork & Migrate

Cluster forking occurs when one or more nodes in a cluster (which had previously agreed to allow a specific joint computing environment to join them) decide to remove the specific computing environment.

When a node decides to remove a joint computing environment from the cluster it belongs to, the other nodes in the cluster have until the start of the next epoch to either support the ejection or ignore it and continue providing computational services as at the beginning of the next epoch the cluster will split.

The nodes that chose to eject the joint computing environment will need to equally divide the costs of creating a new cluster that will continue to support the joint computing environments.

Migration can occur for two reasons: planned migration and forced migration.

Planned Migration

Planned migration occurs when a node decides to exit the network voluntarily. To avoid penalties, the node must announce its exit one epoch ahead of time. The costs of migrating to a new cluster are deducted from the node's self-delegation before the remaining self-delegation is returned. A network constant ensures that the self-delegation is always large enough to cover any migration costs.

Forced Migration

Forced migration occurs when a node is penalized for reasons such as extended downtime or misconduct or when there’s a significant drop in stake. In such cases, the migration costs are shared proportionally by all stakeholders who are collectively responsible for monitoring the node’s performance and stake level. These costs are typically small, so stakeholders will likely see only a minor reduction in their rewards for the epoch in which the migration takes place.