This is a part of ZDT epic: #25820
When a node starts, it creates its MerkleDb database in a temp folder in data/saved/swirlds-tmp. Database directory name is something like 1781560433905119-merkledb-state, i.e. a timestamp plus "merkledb" plus "state" (which used to be a virtual map label, but now is always hard-coded as "state").
With the current ZDT proposal, the new CN process doesn't create its own MerkleDb (from scratch or from a snapshot), but it needs to pick up the instance from the old CN process. To do that, the name of MerkleDb directory should not be random (the timestamp component), but fixed / deterministic.
This is a part of ZDT epic: #25820
When a node starts, it creates its MerkleDb database in a temp folder in
data/saved/swirlds-tmp. Database directory name is something like1781560433905119-merkledb-state, i.e. a timestamp plus "merkledb" plus "state" (which used to be a virtual map label, but now is always hard-coded as "state").With the current ZDT proposal, the new CN process doesn't create its own MerkleDb (from scratch or from a snapshot), but it needs to pick up the instance from the old CN process. To do that, the name of MerkleDb directory should not be random (the timestamp component), but fixed / deterministic.