Struct DiskDb

pub struct DiskDb {
    db_kind: String,
    format_version_in_code: Version,
    network: Network,
    ephemeral: bool,
    db: Arc<DBWithThreadMode<SingleThreaded>>,
}

Wrapper struct to ensure low-level database access goes through the correct API.

rocksdb allows concurrent writes through a shared reference, so database instances are cloneable. When the final clone is dropped, the database is closed.

Correctness

Reading transactions from the database using RocksDB iterators causes hangs. But creating iterators and reading the tip height works fine.

So these hangs are probably caused by holding column family locks to read:

• multiple values, or
• large values.

This bug might be fixed by moving database operations to blocking threads (#2188), so that they don’t block the tokio executor. (Or it might be fixed by future RocksDB upgrades.)

Fields

db_kind: String

The configured database kind for this database.

format_version_in_code: Version

The format version of the running Zebra code.

network: Network

The configured network for this database.

ephemeral: bool

The configured temporary database setting.

If true, the database files are deleted on drop.

db: Arc<DBWithThreadMode<SingleThreaded>>

The shared inner RocksDB database.

RocksDB allows reads and writes via a shared reference.

In SingleThreaded mode, column family changes and Drop require exclusive access.

In MultiThreaded mode, only Drop requires exclusive access.

Implementations

impl DiskDb

pub fn list_cf(&self) -> Result<Vec<String>, Error>

Returns a list of column family names in this database.

impl DiskDb

const IDEAL_OPEN_FILE_LIMIT: u64 = 1_024u64

The ideal open file limit for Zebra

const MIN_OPEN_FILE_LIMIT: u64 = 512u64

The minimum number of open files for Zebra to operate normally. Also used as the default open file limit, when the OS doesn’t tell us how many files we can use.

We want 100+ file descriptors for peers, and 100+ for the database.

On Windows, the default limit is 512 high-level I/O files, and 8192 low-level I/O files: https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/setmaxstdio?view=msvc-160#remarks

const RESERVED_FILE_COUNT: u64 = 48u64

The number of files used internally by Zebra.

Zebra uses file descriptors for OS libraries (10+), polling APIs (10+), stdio (3), and other OS facilities (2+).

const MEMTABLE_RAM_CACHE_MEGABYTES: usize = 128usize

The size of the database memtable RAM cache in megabytes.

https://github.com/facebook/rocksdb/wiki/RocksDB-FAQ#configuration-and-tuning

pub fn print_db_metrics(&self)

Prints rocksdb metrics for each column family along with total database disk size, live data disk size and database memory size.

pub fn try_catch_up_with_primary(&self) -> Result<(), Error>

When called with a secondary DB instance, tries to catch up with the primary DB instance

pub fn zs_forward_range_iter<C, K, V, R>( &self, cf: &C, range: R, ) -> impl Iterator<Item = (K, V)> + '_

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk, R: RangeBounds<K>,

Returns a forward iterator over the items in cf in range.

Holding this iterator open might delay block commit transactions.

pub fn zs_reverse_range_iter<C, K, V, R>( &self, cf: &C, range: R, ) -> impl Iterator<Item = (K, V)> + '_

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk, R: RangeBounds<K>,

Returns a reverse iterator over the items in cf in range.

Holding this iterator open might delay block commit transactions.

fn zs_range_iter_with_direction<C, K, V, R>( &self, cf: &C, range: R, reverse: bool, ) -> impl Iterator<Item = (K, V)> + '_

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk, R: RangeBounds<K>,

Returns an iterator over the items in cf in range.

RocksDB iterators are ordered by increasing key bytes by default. Otherwise, if reverse is true, the iterator is ordered by decreasing key bytes.

Holding this iterator open might delay block commit transactions.

fn zs_iter_opts<R>(range: &R) -> ReadOptions

where R: RangeBounds<Vec<u8>>,

Returns the RocksDB ReadOptions with a lower and upper bound for a range.

fn zs_iter_bounds<R>(range: &R) -> (Option<Vec<u8>>, Option<Vec<u8>>)

where R: RangeBounds<Vec<u8>>,

Returns a lower and upper iterate bounds for a range.

Note: Since upper iterate bounds are always exclusive in RocksDB, this method will increment the upper bound by 1 if the end bound of the provided range is inclusive.

fn zs_iter_mode<R>(range: &R, reverse: bool) -> IteratorMode<'_>

where R: RangeBounds<Vec<u8>>,

Returns the RocksDB iterator “from” mode for range.

RocksDB iterators are ordered by increasing key bytes by default. Otherwise, if reverse is true, the iterator is ordered by decreasing key bytes.

fn construct_column_families( db_options: &Options, path: &Path, column_families_in_code: &[String], ) -> Vec<ColumnFamilyDescriptor>

Build a vector of current column families on the disk and optionally any new column families. Returns an iterable collection of all column families.

pub fn new( config: &Config, db_kind: impl AsRef<str>, format_version_in_code: &Version, network: &Network, column_families_in_code: impl IntoIterator<Item = String>, read_only: bool, ) -> DiskDb

Opens or creates the database at a path based on the kind, major version and network, with the supplied column families, preserving any existing column families, and returns a shared low-level database wrapper.

pub fn db_kind(&self) -> String

Returns the configured database kind for this database.

pub fn format_version_in_code(&self) -> Version

Returns the format version of the running code that created this DiskDb instance in memory.

pub fn major_version(&self) -> u64

Returns the fixed major version for this database.

pub fn network(&self) -> Network

Returns the configured network for this database.

pub fn path(&self) -> &Path

Returns the Path where the files used by this database are located.

fn inner(&self) -> &Arc<DBWithThreadMode<SingleThreaded>>

Returns the low-level rocksdb inner database.

pub fn cf_handle(&self, cf_name: &str) -> Option<ColumnFamilyRef<'_>>

Returns the column family handle for cf_name.

pub(crate) fn write(&self, batch: DiskWriteBatch) -> Result<(), Error>

Writes batch to the database.

pub(crate) fn try_reusing_previous_db_after_major_upgrade( restorable_db_versions: &[u64], format_version_in_code: &Version, config: &Config, db_kind: impl AsRef<str>, network: &Network, )

Tries to reuse an existing db after a major upgrade.

If the current db version belongs to restorable_db_versions, the function moves a previous db to a new path so it can be used again. It does so by merely trying to rename the path corresponding to the db version directly preceding the current version to the path that is used by the current db. If successful, it also deletes the db version file.

fn options() -> Options

Returns the database options for the finalized state database.

fn get_db_open_file_limit(open_file_limit: u64) -> u64

Calculate the database’s share of open_file_limit

fn increase_open_file_limit() -> u64

Increase the open file limit for this process to IDEAL_OPEN_FILE_LIMIT. If that fails, try MIN_OPEN_FILE_LIMIT.

If the current limit is above IDEAL_OPEN_FILE_LIMIT, leaves it unchanged.

Returns the current limit, after any successful increases.

Panics

If the open file limit can not be increased to MIN_OPEN_FILE_LIMIT.

pub(crate) fn shared_database_owners(&self) -> usize

Returns the number of shared instances of this database.

Concurrency

The actual number of owners can be higher or lower than the returned value, because databases can simultaneously be cloned or dropped in other threads.

However, if the number of owners is 1, and the caller has exclusive access, the count can’t increase unless that caller clones the database.

pub(crate) fn shutdown(&mut self, force: bool)

Shut down the database, cleaning up background tasks and ephemeral data.

If force is true, clean up regardless of any shared references. force can cause errors accessing the database from other shared references. It should only be used in debugging or test code, immediately before a manual shutdown.

TODO: make private after the stop height check has moved to the syncer (#3442) move shutting down the database to a blocking thread (#2188)

fn delete_ephemeral(&mut self)

If the database is ephemeral, delete its files.

fn assert_default_cf_is_empty(&self)

Check that the “default” column family is empty.

Panics

If Zebra has a bug where it is storing data in the wrong column family.

Methods from Deref<Target = DBWithThreadMode<SingleThreaded>>

pub fn delete_range_cf_opt<K>( &self, cf: &impl AsColumnFamilyRef, from: K, to: K, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>,

Removes the database entries in the range ["from", "to") using given write options.

pub fn delete_range_cf<K>( &self, cf: &impl AsColumnFamilyRef, from: K, to: K, ) -> Result<(), Error>

where K: AsRef<[u8]>,

Removes the database entries in the range ["from", "to") using default write options.

pub fn write_opt( &self, batch: WriteBatchWithTransaction<false>, writeopts: &WriteOptions, ) -> Result<(), Error>

pub fn write( &self, batch: WriteBatchWithTransaction<false>, ) -> Result<(), Error>

pub fn write_without_wal( &self, batch: WriteBatchWithTransaction<false>, ) -> Result<(), Error>

pub fn path(&self) -> &Path

pub fn flush_wal(&self, sync: bool) -> Result<(), Error>

Flushes the WAL buffer. If sync is set to true, also syncs the data to disk.

pub fn flush_opt(&self, flushopts: &FlushOptions) -> Result<(), Error>

Flushes database memtables to SST files on the disk.

pub fn flush(&self) -> Result<(), Error>

Flushes database memtables to SST files on the disk using default options.

pub fn flush_cf_opt( &self, cf: &impl AsColumnFamilyRef, flushopts: &FlushOptions, ) -> Result<(), Error>

Flushes database memtables to SST files on the disk for a given column family.

pub fn flush_cfs_opt( &self, cfs: &[&impl AsColumnFamilyRef], opts: &FlushOptions, ) -> Result<(), Error>

Flushes multiple column families.

If atomic flush is not enabled, it is equivalent to calling flush_cf multiple times. If atomic flush is enabled, it will flush all column families specified in cfs up to the latest sequence number at the time when flush is requested.

pub fn flush_cf(&self, cf: &impl AsColumnFamilyRef) -> Result<(), Error>

Flushes database memtables to SST files on the disk for a given column family using default options.

pub fn get_opt<K>( &self, key: K, readopts: &ReadOptions, ) -> Result<Option<Vec<u8>>, Error>

where K: AsRef<[u8]>,

Return the bytes associated with a key value with read options. If you only intend to use the vector returned temporarily, consider using get_pinned_opt to avoid unnecessary memory copy.

pub fn get<K>(&self, key: K) -> Result<Option<Vec<u8>>, Error>

where K: AsRef<[u8]>,

Return the bytes associated with a key value. If you only intend to use the vector returned temporarily, consider using get_pinned to avoid unnecessary memory copy.

pub fn get_cf_opt<K>( &self, cf: &impl AsColumnFamilyRef, key: K, readopts: &ReadOptions, ) -> Result<Option<Vec<u8>>, Error>

where K: AsRef<[u8]>,

Return the bytes associated with a key value and the given column family with read options. If you only intend to use the vector returned temporarily, consider using get_pinned_cf_opt to avoid unnecessary memory.

pub fn get_cf<K>( &self, cf: &impl AsColumnFamilyRef, key: K, ) -> Result<Option<Vec<u8>>, Error>

where K: AsRef<[u8]>,

Return the bytes associated with a key value and the given column family. If you only intend to use the vector returned temporarily, consider using get_pinned_cf to avoid unnecessary memory.

pub fn get_pinned_opt<K>( &self, key: K, readopts: &ReadOptions, ) -> Result<Option<DBPinnableSlice<'_>>, Error>

where K: AsRef<[u8]>,

Return the value associated with a key using RocksDB’s PinnableSlice so as to avoid unnecessary memory copy.

pub fn get_pinned<K>( &self, key: K, ) -> Result<Option<DBPinnableSlice<'_>>, Error>

where K: AsRef<[u8]>,

Return the value associated with a key using RocksDB’s PinnableSlice so as to avoid unnecessary memory copy. Similar to get_pinned_opt but leverages default options.

pub fn get_pinned_cf_opt<K>( &self, cf: &impl AsColumnFamilyRef, key: K, readopts: &ReadOptions, ) -> Result<Option<DBPinnableSlice<'_>>, Error>

where K: AsRef<[u8]>,

Return the value associated with a key using RocksDB’s PinnableSlice so as to avoid unnecessary memory copy. Similar to get_pinned_opt but allows specifying ColumnFamily

pub fn get_pinned_cf<K>( &self, cf: &impl AsColumnFamilyRef, key: K, ) -> Result<Option<DBPinnableSlice<'_>>, Error>

where K: AsRef<[u8]>,

Return the value associated with a key using RocksDB’s PinnableSlice so as to avoid unnecessary memory copy. Similar to get_pinned_cf_opt but leverages default options.

pub fn multi_get<K, I>(&self, keys: I) -> Vec<Result<Option<Vec<u8>>, Error>>

where K: AsRef<[u8]>, I: IntoIterator<Item = K>,

Return the values associated with the given keys.

pub fn multi_get_opt<K, I>( &self, keys: I, readopts: &ReadOptions, ) -> Vec<Result<Option<Vec<u8>>, Error>>

where K: AsRef<[u8]>, I: IntoIterator<Item = K>,

Return the values associated with the given keys using read options.

pub fn multi_get_cf<'a, 'b, K, I, W>( &'a self, keys: I, ) -> Vec<Result<Option<Vec<u8>>, Error>>

where 'b: 'a, K: AsRef<[u8]>, I: IntoIterator<Item = (&'b W, K)>, W: 'b + AsColumnFamilyRef,

Return the values associated with the given keys and column families.

pub fn multi_get_cf_opt<'a, 'b, K, I, W>( &'a self, keys: I, readopts: &ReadOptions, ) -> Vec<Result<Option<Vec<u8>>, Error>>

where 'b: 'a, K: AsRef<[u8]>, I: IntoIterator<Item = (&'b W, K)>, W: 'b + AsColumnFamilyRef,

Return the values associated with the given keys and column families using read options.

pub fn batched_multi_get_cf<'a, K, I>( &self, cf: &impl AsColumnFamilyRef, keys: I, sorted_input: bool, ) -> Vec<Result<Option<DBPinnableSlice<'_>>, Error>>

where K: AsRef<[u8]> + 'a + ?Sized, I: IntoIterator<Item = &'a K>,

Return the values associated with the given keys and the specified column family where internally the read requests are processed in batch if block-based table SST format is used. It is a more optimized version of multi_get_cf.

pub fn batched_multi_get_cf_opt<'a, K, I>( &self, cf: &impl AsColumnFamilyRef, keys: I, sorted_input: bool, readopts: &ReadOptions, ) -> Vec<Result<Option<DBPinnableSlice<'_>>, Error>>

where K: AsRef<[u8]> + 'a + ?Sized, I: IntoIterator<Item = &'a K>,

Return the values associated with the given keys and the specified column family where internally the read requests are processed in batch if block-based table SST format is used. It is a more optimized version of multi_get_cf_opt.

pub fn key_may_exist<K>(&self, key: K) -> bool

where K: AsRef<[u8]>,

Returns false if the given key definitely doesn’t exist in the database, otherwise returns true. This function uses default ReadOptions.

pub fn key_may_exist_opt<K>(&self, key: K, readopts: &ReadOptions) -> bool

where K: AsRef<[u8]>,

Returns false if the given key definitely doesn’t exist in the database, otherwise returns true.

pub fn key_may_exist_cf<K>(&self, cf: &impl AsColumnFamilyRef, key: K) -> bool

where K: AsRef<[u8]>,

Returns false if the given key definitely doesn’t exist in the specified column family, otherwise returns true. This function uses default ReadOptions.

pub fn key_may_exist_cf_opt<K>( &self, cf: &impl AsColumnFamilyRef, key: K, readopts: &ReadOptions, ) -> bool

where K: AsRef<[u8]>,

Returns false if the given key definitely doesn’t exist in the specified column family, otherwise returns true.

pub fn key_may_exist_cf_opt_value<K>( &self, cf: &impl AsColumnFamilyRef, key: K, readopts: &ReadOptions, ) -> (bool, Option<CSlice>)

where K: AsRef<[u8]>,

If the key definitely does not exist in the database, then this method returns (false, None), else (true, None) if it may. If the key is found in memory, then it returns (true, Some<CSlice>).

This check is potentially lighter-weight than calling get(). One way to make this lighter weight is to avoid doing any IOs.

pub fn iterator<'a, 'b>( &'a self, mode: IteratorMode<'_>, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

pub fn iterator_opt<'a, 'b>( &'a self, mode: IteratorMode<'_>, readopts: ReadOptions, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

pub fn iterator_cf_opt<'a, 'b>( &'a self, cf_handle: &impl AsColumnFamilyRef, readopts: ReadOptions, mode: IteratorMode<'_>, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

Opens an iterator using the provided ReadOptions. This is used when you want to iterate over a specific ColumnFamily with a modified ReadOptions

pub fn full_iterator<'a, 'b>( &'a self, mode: IteratorMode<'_>, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

Opens an iterator with set_total_order_seek enabled. This must be used to iterate across prefixes when set_memtable_factory has been called with a Hash-based implementation.

pub fn prefix_iterator<'a, 'b, P>( &'a self, prefix: P, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b, P: AsRef<[u8]>,

pub fn iterator_cf<'a, 'b>( &'a self, cf_handle: &impl AsColumnFamilyRef, mode: IteratorMode<'_>, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

pub fn full_iterator_cf<'a, 'b>( &'a self, cf_handle: &impl AsColumnFamilyRef, mode: IteratorMode<'_>, ) -> DBIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

pub fn prefix_iterator_cf<'a, P>( &'a self, cf_handle: &impl AsColumnFamilyRef, prefix: P, ) -> DBIteratorWithThreadMode<'a, DBCommon<T, D>>

where P: AsRef<[u8]>,

pub fn raw_iterator<'a, 'b>( &'a self, ) -> DBRawIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

Opens a raw iterator over the database, using the default read options

pub fn raw_iterator_cf<'a, 'b>( &'a self, cf_handle: &impl AsColumnFamilyRef, ) -> DBRawIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

Opens a raw iterator over the given column family, using the default read options

pub fn raw_iterator_opt<'a, 'b>( &'a self, readopts: ReadOptions, ) -> DBRawIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

Opens a raw iterator over the database, using the given read options

pub fn raw_iterator_cf_opt<'a, 'b>( &'a self, cf_handle: &impl AsColumnFamilyRef, readopts: ReadOptions, ) -> DBRawIteratorWithThreadMode<'b, DBCommon<T, D>>

where 'a: 'b,

Opens a raw iterator over the given column family, using the given read options

pub fn snapshot(&self) -> SnapshotWithThreadMode<'_, DBCommon<T, D>>

pub fn put_opt<K, V>( &self, key: K, value: V, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn put_cf_opt<K, V>( &self, cf: &impl AsColumnFamilyRef, key: K, value: V, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn merge_opt<K, V>( &self, key: K, value: V, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn merge_cf_opt<K, V>( &self, cf: &impl AsColumnFamilyRef, key: K, value: V, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn delete_opt<K>( &self, key: K, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>,

pub fn delete_cf_opt<K>( &self, cf: &impl AsColumnFamilyRef, key: K, writeopts: &WriteOptions, ) -> Result<(), Error>

where K: AsRef<[u8]>,

pub fn put<K, V>(&self, key: K, value: V) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn put_cf<K, V>( &self, cf: &impl AsColumnFamilyRef, key: K, value: V, ) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn merge<K, V>(&self, key: K, value: V) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn merge_cf<K, V>( &self, cf: &impl AsColumnFamilyRef, key: K, value: V, ) -> Result<(), Error>

where K: AsRef<[u8]>, V: AsRef<[u8]>,

pub fn delete<K>(&self, key: K) -> Result<(), Error>

where K: AsRef<[u8]>,

pub fn delete_cf<K>( &self, cf: &impl AsColumnFamilyRef, key: K, ) -> Result<(), Error>

where K: AsRef<[u8]>,

pub fn compact_range<S, E>(&self, start: Option<S>, end: Option<E>)

where S: AsRef<[u8]>, E: AsRef<[u8]>,

Runs a manual compaction on the Range of keys given. This is not likely to be needed for typical usage.

pub fn compact_range_opt<S, E>( &self, start: Option<S>, end: Option<E>, opts: &CompactOptions, )

where S: AsRef<[u8]>, E: AsRef<[u8]>,

Same as compact_range but with custom options.

pub fn compact_range_cf<S, E>( &self, cf: &impl AsColumnFamilyRef, start: Option<S>, end: Option<E>, )

where S: AsRef<[u8]>, E: AsRef<[u8]>,

Runs a manual compaction on the Range of keys given on the given column family. This is not likely to be needed for typical usage.

pub fn compact_range_cf_opt<S, E>( &self, cf: &impl AsColumnFamilyRef, start: Option<S>, end: Option<E>, opts: &CompactOptions, )

where S: AsRef<[u8]>, E: AsRef<[u8]>,

Same as compact_range_cf but with custom options.

pub fn wait_for_compact( &self, opts: &WaitForCompactOptions, ) -> Result<(), Error>

Wait for all flush and compactions jobs to finish. Jobs to wait include the unscheduled (queued, but not scheduled yet).

NOTE: This may also never return if there’s sufficient ongoing writes that keeps flush and compaction going without stopping. The user would have to cease all the writes to DB to make this eventually return in a stable state. The user may also use timeout option in WaitForCompactOptions to make this stop waiting and return when timeout expires.

pub fn set_options(&self, opts: &[(&str, &str)]) -> Result<(), Error>

pub fn set_options_cf( &self, cf: &impl AsColumnFamilyRef, opts: &[(&str, &str)], ) -> Result<(), Error>

pub fn property_value( &self, name: impl CStrLike, ) -> Result<Option<String>, Error>

Retrieves a RocksDB property by name.

Full list of properties could be find here.

pub fn property_value_cf( &self, cf: &impl AsColumnFamilyRef, name: impl CStrLike, ) -> Result<Option<String>, Error>

Retrieves a RocksDB property by name, for a specific column family.

Full list of properties could be find here.

pub fn property_int_value( &self, name: impl CStrLike, ) -> Result<Option<u64>, Error>

Retrieves a RocksDB property and casts it to an integer.

Full list of properties that return int values could be find here.

pub fn property_int_value_cf( &self, cf: &impl AsColumnFamilyRef, name: impl CStrLike, ) -> Result<Option<u64>, Error>

Retrieves a RocksDB property for a specific column family and casts it to an integer.

Full list of properties that return int values could be find here.

pub fn latest_sequence_number(&self) -> u64

The sequence number of the most recent transaction.

pub fn get_updates_since(&self, seq_number: u64) -> Result<DBWALIterator, Error>

Iterate over batches of write operations since a given sequence.

Produce an iterator that will provide the batches of write operations that have occurred since the given sequence (see latest_sequence_number()). Use the provided iterator to retrieve each (u64, WriteBatch) tuple, and then gather the individual puts and deletes using the WriteBatch::iterate() function.

Calling get_updates_since() with a sequence number that is out of bounds will return an error.

pub fn try_catch_up_with_primary(&self) -> Result<(), Error>

Tries to catch up with the primary by reading as much as possible from the log files.

pub fn ingest_external_file<P>(&self, paths: Vec<P>) -> Result<(), Error>

where P: AsRef<Path>,

Loads a list of external SST files created with SstFileWriter into the DB with default opts

pub fn ingest_external_file_opts<P>( &self, opts: &IngestExternalFileOptions, paths: Vec<P>, ) -> Result<(), Error>

where P: AsRef<Path>,

Loads a list of external SST files created with SstFileWriter into the DB

pub fn ingest_external_file_cf<P>( &self, cf: &impl AsColumnFamilyRef, paths: Vec<P>, ) -> Result<(), Error>

where P: AsRef<Path>,

Loads a list of external SST files created with SstFileWriter into the DB for given Column Family with default opts

pub fn ingest_external_file_cf_opts<P>( &self, cf: &impl AsColumnFamilyRef, opts: &IngestExternalFileOptions, paths: Vec<P>, ) -> Result<(), Error>

where P: AsRef<Path>,

Loads a list of external SST files created with SstFileWriter into the DB for given Column Family

pub fn get_column_family_metadata(&self) -> ColumnFamilyMetaData

Obtains the LSM-tree meta data of the default column family of the DB

pub fn get_column_family_metadata_cf( &self, cf: &impl AsColumnFamilyRef, ) -> ColumnFamilyMetaData

Obtains the LSM-tree meta data of the specified column family of the DB

pub fn live_files(&self) -> Result<Vec<LiveFile>, Error>

Returns a list of all table files with their level, start key and end key

pub fn delete_file_in_range<K>(&self, from: K, to: K) -> Result<(), Error>

where K: AsRef<[u8]>,

Delete sst files whose keys are entirely in the given range.

Could leave some keys in the range which are in files which are not entirely in the range.

Note: L0 files are left regardless of whether they’re in the range.

SnapshotWithThreadModes before the delete might not see the data in the given range.

pub fn delete_file_in_range_cf<K>( &self, cf: &impl AsColumnFamilyRef, from: K, to: K, ) -> Result<(), Error>

where K: AsRef<[u8]>,

Same as delete_file_in_range but only for specific column family

pub fn cancel_all_background_work(&self, wait: bool)

Request stopping background work, if wait is true wait until it’s done.

pub fn cf_handle(&self, name: &str) -> Option<&ColumnFamily>

Returns the underlying column family handle

pub fn create_cf<N>(&self, name: N, opts: &Options) -> Result<(), Error>

where N: AsRef<str>,

Creates column family with given name and options

pub fn drop_cf(&self, name: &str) -> Result<(), Error>

Drops the column family with the given name by internally locking the inner column family map. This avoids needing &mut self reference

pub fn cf_handle(&self, name: &str) -> Option<Arc<BoundColumnFamily<'_>>>

Returns the underlying column family handle

pub fn transaction( &self, ) -> Transaction<'_, DBCommon<T, OptimisticTransactionDBInner>>

Creates a transaction with default options.

pub fn transaction_opt( &self, writeopts: &WriteOptions, otxn_opts: &OptimisticTransactionOptions, ) -> Transaction<'_, DBCommon<T, OptimisticTransactionDBInner>>

Creates a transaction with default options.

pub fn write_opt( &self, batch: WriteBatchWithTransaction<true>, writeopts: &WriteOptions, ) -> Result<(), Error>

pub fn write(&self, batch: WriteBatchWithTransaction<true>) -> Result<(), Error>

pub fn write_without_wal( &self, batch: WriteBatchWithTransaction<true>, ) -> Result<(), Error>

Trait Implementations

impl Clone for DiskDb

fn clone(&self) -> DiskDb

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for DiskDb

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Drop for DiskDb

fn drop(&mut self)

Executes the destructor for this type.

impl PartialEq for DiskDb

fn eq(&self, other: &Self) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl ReadDisk for DiskDb

Deprecation

These impls should not be used in new code, use TypedColumnFamily instead.

fn zs_is_empty<C>(&self, cf: &C) -> bool

where C: AsColumnFamilyRef,

Returns true if a rocksdb column family cf does not contain any entries.

fn zs_get<C, K, V>(&self, cf: &C, key: &K) -> Option<V>

where C: AsColumnFamilyRef, K: IntoDisk, V: FromDisk,

Returns the value for key in the rocksdb column family cf, if present.

fn zs_contains<C, K>(&self, cf: &C, key: &K) -> bool

where C: AsColumnFamilyRef, K: IntoDisk,

Check if a rocksdb column family cf contains the serialized form of key.

fn zs_first_key_value<C, K, V>(&self, cf: &C) -> Option<(K, V)>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk,

Returns the lowest key in cf, and the corresponding value.

fn zs_last_key_value<C, K, V>(&self, cf: &C) -> Option<(K, V)>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk,

Returns the highest key in cf, and the corresponding value.

fn zs_next_key_value_from<C, K, V>( &self, cf: &C, lower_bound: &K, ) -> Option<(K, V)>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk,

Returns the first key greater than or equal to lower_bound in cf, and the corresponding value.

fn zs_next_key_value_strictly_after<C, K, V>( &self, cf: &C, lower_bound: &K, ) -> Option<(K, V)>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk,

Returns the first key strictly greater than lower_bound in cf, and the corresponding value.

fn zs_prev_key_value_back_from<C, K, V>( &self, cf: &C, upper_bound: &K, ) -> Option<(K, V)>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk,

Returns the first key less than or equal to upper_bound in cf, and the corresponding value.

fn zs_prev_key_value_strictly_before<C, K, V>( &self, cf: &C, upper_bound: &K, ) -> Option<(K, V)>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk, V: FromDisk,

Returns the first key strictly less than upper_bound in cf, and the corresponding value.

fn zs_items_in_range_ordered<C, K, V, R>( &self, cf: &C, range: R, ) -> BTreeMap<K, V>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk + Ord, V: FromDisk, R: RangeBounds<K>,

Returns the keys and values in cf in range, in an ordered BTreeMap.

fn zs_items_in_range_unordered<C, K, V, R>( &self, cf: &C, range: R, ) -> HashMap<K, V>

where C: AsColumnFamilyRef, K: IntoDisk + FromDisk + Eq + Hash, V: FromDisk, R: RangeBounds<K>,

Returns the keys and values in cf in range, in an unordered HashMap.

impl Deref for DiskDb

type Target = DBCommon<SingleThreaded, DBWithThreadModeInner>

The resulting type after dereferencing.

fn deref(&self) -> &Self::Target

Dereferences the value.

impl Eq for DiskDb