Problem
Execution node restarts cause ~10 minutes of downtime while the mtrie loads from disk. During HCU, all execution nodes restart at the same height, so no execution node is available to execute the next block until the mtrie reload completes, causing sealing downtime.
The ledger (mtrie) is used to read register values from execution state and to generate proofs for block execution. Moving the ledger to a standalone service would allow execution nodes to restart faster, since the ledger service remains available. This would reduce sealing downtime during HCU.
Solution
- Create a ledger service that implements the ledger interface .
- Implement RPC communication between the execution node process and the ledger service.
Problem
Execution node restarts cause ~10 minutes of downtime while the mtrie loads from disk. During HCU, all execution nodes restart at the same height, so no execution node is available to execute the next block until the mtrie reload completes, causing sealing downtime.
The ledger (mtrie) is used to read register values from execution state and to generate proofs for block execution. Moving the ledger to a standalone service would allow execution nodes to restart faster, since the ledger service remains available. This would reduce sealing downtime during HCU.
Solution