You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, Citrea nodes (Sequencer / Full node / Prover) run in archive mode. Meaning that they store all block data ever since genesis all the way to the tip of the chain.
We'd like to implement pruning so that we can be selective on how much data we retain based on specific configuration parameters.
Based on Reth's pruning documentation, we'd like to implement a similar approach.
There are multiple areas which we should look into:
EVM
ProverStorageManager
LedgerDB tables.
This is a parent issue which outlines the following tasks that need to be done:
Research each component and create child issues define what needs to be done
Given that our underlying database is RocksDB to which we write in a blocking fashion, running the pruner to delete records is going to be a time consuming process. Therefore, consider running the pruner in the async pool and using spawn_blocking for invoking writes or running the whole component in the blocking pool.
The text was updated successfully, but these errors were encountered:
Currently, Citrea nodes (Sequencer / Full node / Prover) run in archive mode. Meaning that they store all block data ever since genesis all the way to the tip of the chain.
We'd like to implement pruning so that we can be selective on how much data we retain based on specific configuration parameters.
Based on Reth's pruning documentation, we'd like to implement a similar approach.
There are multiple areas which we should look into:
EVM
ProverStorageManager
LedgerDB
tables.This is a parent issue which outlines the following tasks that need to be done:
Concerns
Given that our underlying database is RocksDB to which we write in a blocking fashion, running the pruner to delete records is going to be a time consuming process. Therefore, consider running the pruner in the async pool and using
spawn_blocking
for invoking writes or running the whole component in the blocking pool.The text was updated successfully, but these errors were encountered: