This page would describe ideas and proposals on how to implement addition and removal of nodes in an XC cluster. When we consider the problem of node addition and removal, we have to first address
- Node identification using some other means rather than numbers
- Table redistribution and tuple relocation will be required when a new node is added.
- How would the new node get the catalog? One idea was to copy it from the coordinator using synchronous streaming replication.
- Whether the node addition/removal be Online/Offline?