Struct tikv::storage::txn::commands::prewrite::Prewriter

struct Prewriter<K: PrewriteKind> {
    kind: K,
    mutations: Vec<K::Mutation>,
    primary: Vec<u8>,
    start_ts: TimeStamp,
    lock_ttl: u64,
    txn_size: u64,
    min_commit_ts: TimeStamp,
    max_commit_ts: TimeStamp,
    secondary_keys: Option<Vec<Vec<u8>>>,
    old_values: OldValues,
    try_one_pc: bool,
    ctx: Context,
}

Handles both kinds of prewrite (K statically indicates either optimistic or pessimistic).

Fields

kind: K

mutations: Vec<K::Mutation>

primary: Vec<u8>

start_ts: TimeStamp

lock_ttl: u64

txn_size: u64

min_commit_ts: TimeStamp

max_commit_ts: TimeStamp

secondary_keys: Option<Vec<Vec<u8>>>

old_values: OldValues

try_one_pc: bool

ctx: Context

Implementations

impl<K: PrewriteKind> Prewriter<K>

fn process_write(
    self,
    snapshot: impl Snapshot,
    context: WriteContext<'_, impl LockManager>
) -> Result<WriteResult>

Entry point for handling a prewrite by Prewriter.

fn check_max_ts_synced(&self, snapshot: &impl Snapshot) -> Result<()>

fn prewrite(
    &mut self,
    txn: &mut MvccTxn,
    reader: &mut SnapshotReader<impl Snapshot>,
    extra_op: ExtraOp
) -> Result<(Vec<Result<(), StorageError>>, TimeStamp)>

The core part of the prewrite action. In the abstract, this method iterates over the mutations in the prewrite and prewrites each one. It keeps track of any locks encountered and (if it’s an async commit transaction) the min_commit_ts, these are returned by the method.

fn write_result(
    self,
    locks: Vec<Result<(), StorageError>>,
    txn: MvccTxn,
    final_min_commit_ts: TimeStamp,
    rows: usize,
    async_apply_prewrite: bool,
    lock_manager: &impl LockManager
) -> WriteResult

Prepare a WriteResult object from the results of executing the prewrite.