Struct ClickHouseConfig 

pub struct ClickHouseConfig {

    pub url: String,
    pub table: String,
    pub database: Option<String>,
    pub username: Option<String>,
    pub password: Option<String>,
    pub max_batch_rows: usize,
    pub max_retries: u32,
    pub base_retry_delay: Duration,
    pub max_retry_delay: Duration,
    pub connect_timeout: Duration,
    pub read_timeout: Duration,
    pub gzip: bool,
    pub workers: usize,
    pub spill_dir: Option<PathBuf>,
}

Configuration for the ClickHouse writer.

Fields

url: String

ClickHouse HTTP endpoint (e.g. http://localhost:8123 or https://...).

HTTPS requires the tls feature (enabled by default).

table: String

Target table name for INSERT INTO ... FORMAT ArrowStream.

database: Option<String>

Optional target database. When set, sent via X-ClickHouse-Database.

username: Option<String>

Optional ClickHouse username. When set, sent via X-ClickHouse-User.

password: Option<String>

Optional ClickHouse password/key. When set, sent via X-ClickHouse-Key.

max_batch_rows: usize

Advisory target batch size in rows.

This field is an advisory hint for upstream batching code (such as CollectArrowBridge or TelemetryPipeline) that decide when to flush a partially-filled Arrow batch toward the writer. The writer itself does not split, merge, or reject batches based on this value — whatever RecordBatch is handed to ClickHouseWriter::send is serialized and sent as-is.

Keep this in sync with the batch_size you pass to CollectArrowBridge::new, or read it back from the config to drive that call.

max_retries: u32

Maximum number of retry attempts per batch.

base_retry_delay: Duration

Base delay for exponential backoff (default 100 ms).

Retry N sleeps for base_retry_delay x 2^(N-1) plus jitter.

max_retry_delay: Duration

Maximum backoff cap (default 10 s).

The computed delay is clamped to this value before jitter is applied.

connect_timeout: Duration

TCP connect timeout (default 10 s).

read_timeout: Duration

HTTP read timeout per request (default 60 s).

gzip: bool

If true and the gzip feature is enabled, compress request bodies.

Has no effect when the gzip feature is disabled.

workers: usize

Number of background writer threads draining the channel in parallel.

Defaults to 1 (single-threaded writer preserving submission order). Values greater than 1 allow multiple concurrent HTTP requests to ClickHouse, which is useful when per-batch request latency is the bottleneck. Ordering between batches is not preserved with workers > 1; if downstream consumers rely on the submission order, keep this at 1.

spill_dir: Option<PathBuf>

Optional directory where batches that fail all retries are persisted to disk as Arrow IPC files.

Requires the spill feature. When set, a batch that exhausts every retry is written to spill_dir/batch-{unix_nanos}-{pid}-{seq}.arrows instead of being silently dropped. Replay from disk is not automatic and is left to the operator (e.g. a separate ingest job).

When the feature is disabled this field is ignored.

Implementations

impl ClickHouseConfig

pub fn new(url: impl Into<String>, table: impl Into<String>) -> Self

Create a config with the given endpoint URL and table name.

All other settings use sensible defaults and can be overridden with the builder-style methods below.

pub fn with_max_batch_rows(self, max_batch_rows: usize) -> Self

Override the advisory batch-row target (see max_batch_rows).

pub fn with_max_retries(self, max_retries: u32) -> Self

Override the maximum retry count.

pub fn with_base_retry_delay(self, base_retry_delay: Duration) -> Self

Override the base retry delay.

pub fn with_max_retry_delay(self, max_retry_delay: Duration) -> Self

Override the maximum retry delay.

pub fn with_auth( self, username: impl Into<String>, password: impl Into<String>, ) -> Self

Set ClickHouse authentication credentials (sent via X-ClickHouse-User / X-ClickHouse-Key headers).

pub fn with_database(self, database: impl Into<String>) -> Self

Set the target database (sent via X-ClickHouse-Database).

pub fn with_timeouts(self, connect: Duration, read: Duration) -> Self

Set TCP connect and HTTP read timeouts.

pub fn with_gzip(self, enabled: bool) -> Self

Enable or disable gzip request-body compression.

Requires the gzip Cargo feature to have an effect.

pub fn with_workers(self, workers: usize) -> Self

Override the number of concurrent background writer threads.

See workers for caveats about ordering. Values less than 1 are clamped to 1.

pub fn with_spill_dir(self, dir: impl Into<PathBuf>) -> Self

Enable on-disk spill for undelivered batches.

Requires the spill Cargo feature to have an effect. The directory is created on first spill if it does not already exist.

Trait Implementations

impl Clone for ClickHouseConfig

fn clone(&self) -> ClickHouseConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ClickHouseConfig

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

Formats the value using the given formatter.

impl Default for ClickHouseConfig

fn default() -> Self

Returns the “default value” for a type.