forknet: Drop unneeded columns in the fork-network command

core/store/src/db/rocksdb.rs

@@ -677,13 +677,23 @@
                 })
                 .collect::<Vec<_>>();
             if !values.is_empty() {
                 // TODO(mina86): Once const_str_from_utf8 is stabilised we might
                 // be able convert this runtime UTF-8 validation into const.
                 let stat_name = prop_name.to_str().unwrap();
                 result.data.push((stat_name.to_string(), values));
             }
         }
     }
+
+    /// Deletes all data in `cols`. This should also remove all sst files
+    pub fn clear_cols(&mut self, cols: &[DBCol]) -> anyhow::Result<()> {
+        for col in cols {
+            self.db
+                .drop_cf(col_name(*col))
+                .with_context(|| format!("failed to drop column family {:?}", col,))?;
+        }
+        Ok(())
+    }
 }
 
 impl Drop for RocksDB {

core/store/src/lib.rs

@@ -65,7 +65,8 @@ pub mod trie;
 
 pub use crate::config::{Mode, StoreConfig};
 pub use crate::opener::{
-    checkpoint_hot_storage_and_cleanup_columns, StoreMigrator, StoreOpener, StoreOpenerError,
+    checkpoint_hot_storage_and_cleanup_columns, clear_columns, StoreMigrator, StoreOpener,
+    StoreOpenerError,
 };
 
 /// Specifies temperature of a storage.

core/store/src/opener.rs

@@ -252,14 +252,10 @@ impl<'a> StoreOpener<'a> {
         self.open_in_mode(Mode::ReadWrite)
     }
 
-    /// Opens the RocksDB database(s) for hot and cold (if configured) storages.
-    ///
-    /// When opening in read-only mode, verifies that the database version is
-    /// what the node expects and fails if it isn’t.  If database doesn’t exist,
-    /// creates a new one unless mode is [`Mode::ReadWriteExisting`].  On the
-    /// other hand, if mode is [`Mode::Create`], fails if the database already
-    /// exists.
-    pub fn open_in_mode(&self, mode: Mode) -> Result<crate::NodeStorage, StoreOpenerError> {
+    fn open_dbs(
+        &self,
+        mode: Mode,
+    ) -> Result<(RocksDB, Snapshot, Option<RocksDB>, Snapshot), StoreOpenerError> {
         {
             let hot_path = self.hot.path.display().to_string();
             let cold_path = match &self.cold {
@@ -290,7 +286,18 @@ impl<'a> StoreOpener<'a> {
             .map(|cold| cold.open(mode, DB_VERSION))
             .transpose()?
             .map(|(db, _)| db);
+        Ok((hot_db, hot_snapshot, cold_db, cold_snapshot))
+    }
 
+    /// Opens the RocksDB database(s) for hot and cold (if configured) storages.
+    ///
+    /// When opening in read-only mode, verifies that the database version is
+    /// what the node expects and fails if it isn’t.  If database doesn’t exist,
+    /// creates a new one unless mode is [`Mode::ReadWriteExisting`].  On the
+    /// other hand, if mode is [`Mode::Create`], fails if the database already
+    /// exists.
+    pub fn open_in_mode(&self, mode: Mode) -> Result<crate::NodeStorage, StoreOpenerError> {
+        let (hot_db, hot_snapshot, cold_db, cold_snapshot) = self.open_dbs(mode)?;
         let storage = NodeStorage::from_rocksdb(hot_db, cold_db);
 
         hot_snapshot.remove()?;
@@ -636,6 +643,36 @@ pub fn checkpoint_hot_storage_and_cleanup_columns(
     Ok(node_storage)
 }
 
+/// Deletes all data in the columns in `cols` from the rocksdb data
+/// dir in `home_dir`. This actually removes all the sst files rather than
+/// just logically deleting all keys with a transaction, which would
+/// only give space savings after a compaction.  This is meant to be
+/// used only in tools where certain large columns don't need to be kept around.
+///
+/// For example, when preparing a database for use in a forknet
+/// with the fork-network tool, we only need the state and
+/// flat state, and a few other small columns. So getting rid of
+/// everything else saves quite a bit on the disk space needed for each node.
+pub fn clear_columns<'a>(
+    home_dir: &std::path::Path,
+    config: &StoreConfig,
+    archival_config: Option<ArchivalConfig<'a>>,
+    cols: &[DBCol],
+) -> anyhow::Result<()> {
+    let opener = StoreOpener::new(home_dir, config, archival_config);
+    let (mut hot_db, _hot_snapshot, cold_db, _cold_snapshot) =
+        opener.open_dbs(Mode::ReadWriteExisting)?;
+    hot_db.clear_cols(cols)?;
+    if let Some(mut cold) = cold_db {
+        cold.clear_cols(cols)?;
+    }
+    drop(hot_db);
+    // Here we call open_dbs() to recreate the dropped columns, which should now be empty.
+    let (_hot_db, _hot_snapshot, _cold_db, _cold_snapshot) =
+        opener.open_dbs(Mode::ReadWriteExisting)?;
+    Ok(())
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;

tools/fork-network/src/cli.rs

@@ -142,6 +142,13 @@ fn make_state_roots_key(shard_id: ShardId) -> Vec<u8> {
     format!("{FORKED_ROOTS_KEY_PREFIX}{shard_id}").into_bytes()
 }
 
+/// The minimum set of columns that will be needed to start a node after the `finalize` command runs
+const COLUMNS_TO_KEEP: &[DBCol] = &[DBCol::DbVersion, DBCol::Misc, DBCol::State, DBCol::FlatState];
+
+/// Extra columns needed in the setup before the `finalize` command
+const SETUP_COLUMNS_TO_KEEP: &[DBCol] =
+    &[DBCol::EpochInfo, DBCol::FlatStorageStatus, DBCol::ChunkExtra];
+
 #[derive(clap::Parser)]
 struct ResetCmd;
 
@@ -265,7 +272,7 @@ impl ForkNetworkCommand {
     // Snapshots the DB.
     // Determines parameters that will be used to initialize the new chain.
     // After this completes, almost every DB column can be removed, however this command doesn't delete anything itself.
-    fn init(&self, near_config: &mut NearConfig, home_dir: &Path) -> anyhow::Result<()> {
+    fn write_fork_info(&self, near_config: &mut NearConfig, home_dir: &Path) -> anyhow::Result<()> {
         // Open storage with migration
         let storage = open_storage(&home_dir, near_config).unwrap();
         let store = storage.get_hot_store();
@@ -337,6 +344,24 @@ impl ForkNetworkCommand {
         Ok(())
     }
 
+    fn init(&self, near_config: &mut NearConfig, home_dir: &Path) -> anyhow::Result<()> {
+        self.write_fork_info(near_config, home_dir)?;
+        let mut unwanted_cols = Vec::new();
+        for col in DBCol::iter() {
+            if !COLUMNS_TO_KEEP.contains(&col) && !SETUP_COLUMNS_TO_KEEP.contains(&col) {
+                unwanted_cols.push(col);
+            }
+        }
+        near_store::clear_columns(
+            home_dir,
+            &near_config.config.store,
+            near_config.config.archival_config(),
+            &unwanted_cols,
+        )
+        .context("failed deleting unwanted columns")?;
+        Ok(())
+    }
+
     /// Creates a DB snapshot, then
     /// Updates the state to ensure every account has a full access key that is known to us.
     fn amend_access_keys(
@@ -458,19 +483,20 @@ impl ForkNetworkCommand {
 
     /// Deletes DB columns that are not needed in the new chain.
     fn finalize(&self, near_config: &mut NearConfig, home_dir: &Path) -> anyhow::Result<()> {
-        // Open storage with migration
-        let storage = open_storage(&home_dir, near_config).unwrap();
-        let store = storage.get_hot_store();
-
         tracing::info!("Delete unneeded columns in the original DB");
-        let mut update = store.store_update();
+        let mut unwanted_cols = Vec::new();
         for col in DBCol::iter() {
-            match col {
-                DBCol::DbVersion | DBCol::Misc | DBCol::State | DBCol::FlatState => {}
-                _ => update.delete_all(col),
+            if !COLUMNS_TO_KEEP.contains(&col) {
+                unwanted_cols.push(col);
             }
         }
-        update.commit()?;
+        near_store::clear_columns(
+            home_dir,
+            &near_config.config.store,
+            near_config.config.archival_config(),
+            &unwanted_cols,
+        )
+        .context("failed deleting unwanted columns")?;
         Ok(())
     }