Struct ShardedWritePoolConfig 

pub struct ShardedWritePoolConfig {
    pub shards: usize,
    pub per_shard_buffer: usize,
    pub write_loop: WriteLoopConfig,
}

Tunables for run_sharded_write_pool.

shards = 1 is the legacy single-worker layout and incurs no dispatcher overhead. shards > 1 spawns a dispatcher that hashes each sample’s pv_name to a fixed shard (so any one PV’s samples always land in the same per-shard write_loop, preserving timestamp ordering) and N parallel shard workers — different PVs can append concurrently instead of queueing behind a single task. Sites with many active PVs and a fast STS should set this to min(num_cpus, expected concurrent slow appends).

Fields

shards: usize

Number of parallel shard workers. Must be ≥ 1; 1 is the degenerate case (no dispatcher, behaves identically to a bare write_loop_with_config).

per_shard_buffer: usize

mpsc capacity of each shard’s input channel. The dispatcher try_sends into this; when full, the OFFENDING SHARD’s drop is recorded under archiver_dispatcher_shard_overflow_drops_total{shard=N} and the sample’s per-PV buffer_overflow_drops counter is bumped. Other shards keep flowing — that’s the per-shard isolation guarantee.

write_loop: WriteLoopConfig

Per-worker config (timeouts, flush period). Cloned into each shard.

Trait Implementations

impl Clone for ShardedWritePoolConfig

fn clone(&self) -> ShardedWritePoolConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ShardedWritePoolConfig

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

Formats the value using the given formatter.

impl Default for ShardedWritePoolConfig

fn default() -> Self

Returns the “default value” for a type.