risingwavelabs/risingwave · error · ConnectorError

failed to start the CDC connector before the timeout…

Error message

failed to start the CDC connector before the timeout (default: 60s).
HINT: You can increase the timeout by:
- Alter the running source: ALTER SOURCE <source_name> CONNECTOR WITH (cdc.source.wait.streaming.start.timeout = '<larger_value>');
- Set the session variable and recreate the source: SET cdc_source_wait_streaming_start_timeout = <larger_value>;

What it means

The CDC reader waits for the upstream connector (Debezium engine) to reach a streaming/running state within a configurable timeout (default 60s). If the connector is still not initialized when the deadline passes, the reader bails with this error and the HINT describing how to raise `cdc_source_wait_streaming_start_timeout`.

Solutions

1. Increase the timeout: `ALTER SOURCE <source_name> CONNECTOR WITH (cdc.source.wait.streaming.start.timeout = '300');` (per the HINT).
2. Or session-then-recreate: `SET cdc_source_wait_streaming_start_timeout = 300;` and recreate the source.
3. Check upstream connector logs (.risingwave/log, connector errors) for repeated failures — a real error loops forever, not just slowly.
4. Fix the root cause (network, credentials, upstream DB load) so the connector starts promptly.

Example fix

// before
-- source times out with default 60s
CREATE SOURCE ... WITH (connector='mysql-cdc', ...);
// after
ALTER SOURCE my_source CONNECTOR WITH (cdc.source.wait.streaming.start.timeout = '300');

-- Pre-check connectivity and snapshot size before creating the source:
SELECT COUNT(*) FROM my_table; -- if huge, raise the timeout preemptively
-- Test reachability from RW host: mysql -h <host> -P <port> -u <user> -p

match create_cdc_source().await {
    Err(e) if e.to_string().contains("failed to start the CDC connector before the timeout") => {
        // raise cdc_source_wait_streaming_start_timeout and retry
    }
    r => r,
}

When it happens

Trigger: Source creation or recovery where the underlying Debezium connector takes longer than the timeout to start: very large snapshot in progress, slow initial connection to the upstream DB, broker/registry latency, or the connector erroring and retrying repeatedly.

Common situations: Huge MySQL tables making snapshot take minutes; network latency or DNS slowness to the upstream database; upstream DB connection pool exhaustion; running through slow proxies or Kafka Connect in some deployments.

Understand the failure class

Background: Request timed out: what client-side request timeouts mean across libraries (Request timed out, TIMED_OUT, APITimeoutError) — this error's family across 39 libraries.

• Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

Related errors

• Debezium Mongo needs a `_id` column in table
• source validation failed
• unsupported encoding for DEBEZIUM_MONGO format
• all request confluent registry all timeout
• all senders are dropped

AI-assisted analysis of risingwavelabs/risingwave@6469eb736d (2026-09-11). Data as JSON: /api/errors/90385572c9765e50. Report an issue: GitHub.

Appendix: source

Thrown at src/connector/src/source/cdc/source/reader.rs:172

            })
        });

        // wait for the handshake message
        if let Some(res) = rx.recv().await {
            let resp: GetEventStreamResponse = res?;
            let inited = match resp.control {
                Some(info) => info.handshake_ok,
                None => {
                    tracing::error!(
                        ?source_id,
                        "did not receive the CDC handshake message: {:?}",
                        resp
                    );
                    false
                }
            };
            if !inited {
                bail!(
                    "failed to start the CDC connector before the timeout (default: 60s).\n\
                    HINT: You can increase the timeout by:\n\
                    - Alter the running source: ALTER SOURCE <source_name> CONNECTOR WITH (cdc.source.wait.streaming.start.timeout = '<larger_value>');\n\
                    - Set the session variable and recreate the source: SET cdc_source_wait_streaming_start_timeout = <larger_value>;"
                );
            }
        }
        tracing::info!(?source_id, "CDC connector started");

        let instance = match T::source_type() {
            CdcSourceType::Mysql
            | CdcSourceType::Postgres
            | CdcSourceType::Mongodb
            | CdcSourceType::SqlServer => Self {
                source_id: split.split_id() as u64,
                start_offset: split.start_offset().clone(),
                server_addr: None,
                conn_props,

View on GitHub (pinned to 6469eb736d)