risingwavelabs/risingwave · error · ConnectorError

None is returned by `SELECT sys.fn_cdc_get_max_lsn()`, pleas

Error message

None is returned by `SELECT sys.fn_cdc_get_max_lsn()`, please ensure Sql Server Agent is running.

What it means

The query `SELECT sys.fn_cdc_get_max_lsn()` returned a row but the cell was NULL, i.e. CDC capture jobs have never produced a max LSN. `fn_cdc_get_max_lsn` returns NULL until the SQL Server Agent capture job has populated change tables, so this error points at an environment/agent problem rather than code.

Source

Thrown at src/connector/src/source/cdc/external/sql_server.rs:258

                assert_eq!(
                    bytes.len(),
                    10,
                    "sys.fn_cdc_get_max_lsn() should return a 10 bytes array."
                );
                for byte in &bytes[0..4] {
                    hex_string.push_str(&format!("{:02x}", byte));
                }
                hex_string.push(':');
                for byte in &bytes[4..8] {
                    hex_string.push_str(&format!("{:02x}", byte));
                }
                hex_string.push(':');
                for byte in &bytes[8..10] {
                    hex_string.push_str(&format!("{:02x}", byte));
                }
                hex_string
            }
            None => bail!(
                "None is returned by `SELECT sys.fn_cdc_get_max_lsn()`, please ensure Sql Server Agent is running."
            ),
        };

        tracing::debug!("current max_lsn: {}", max_lsn);

        Ok(CdcOffset::SqlServer(SqlServerOffset {
            change_lsn: max_lsn,
            commit_lsn: MAX_COMMIT_LSN.into(),
        }))
    }

    fn snapshot_read(
        &self,
        table_name: SchemaTableName,
        start_pk: Option<OwnedRow>,
        primary_keys: Vec<String>,
        limit: u32,

View on GitHub (pinned to 6469eb736d)

Solutions

  1. Start SQL Server Agent: `EXEC master.dbo.xp_servicecontrol N'START', N'SQLServerAGENT'` or start the service on the host/container.
  2. Check that the CDC capture and cleanup jobs exist and are running (msdb.dbo.cdc_jobs; sqlagent job history).
  3. Confirm sys.fn_cdc_is_cdc_enabled(DB_ID()) and sys.fn_cdc_has_work_completed return expected values; generate a small change to trigger capture.
  4. If Agent cannot run (e.g. some editions), consider CDC alternatives (CT/Debezium setup) — RisingWave requires a resolvable max LSN.

Example fix

// before
None => bail!("None is returned by `SELECT sys.fn_cdc_get_max_lsn()`, please ensure Sql Server Agent is running."),
// after (user-side check before starting the source)
-- SELECT sys.fn_cdc_get_max_lsn(); -- must be non-NULL before creating the CDC source
Defensive patterns

Strategy: validation

Validate before calling

-- Verify Agent + capture job before starting the source:
SELECT sys.fn_cdc_get_max_lsn(); -- must be non-NULL
SELECT name, enabled FROM msdb.dbo.cdc_jobs WHERE database_id = DB_ID();

Try / catch

match row.try_get::<&[u8], usize>(0) {
    Ok(Some(bytes)) => bytes,
    Ok(None) | Err(_) => return Err(anyhow!("max LSN is NULL; start SQL Server Agent and enable CDC")),
}

Prevention

When it happens

Trigger: CDC enabled on table metadata but SQL Server Agent not running (common on containers, Linux installs, or localdb), or capture job stopped/failed right after enabling CDC, when `current_cdc_offset` is called to compute the start offset.

Common situations: Docker mssql image without Agent started; Agent license/edition limitations; capture job failing due to missing sysadmin; freshly enabled CDC where no transaction has been captured yet.

Understand the failure class

Background: EmptyResultError / "no results found": when an API or scraper succeeds but returns zero rows — this error's family across 9 libraries.

Related errors

AI-assisted analysis of risingwavelabs/risingwave@6469eb736d (2026-09-11). Data as JSON: /api/errors/6b57c2fde59e3a31. Report an issue: GitHub.