risingwavelabs/risingwave · error · SinkError

please set only one of the 'index_column' or 'index' propert

Error message

please set only one of the 'index_column' or 'index' properties.

What it means

`validate_config` enforces that an Elasticsearch/OpenSearch sink specifies its target index in exactly one way: either via the static `index` property or via the `index_column` property (a schema column whose values become the per-record index). Specifying both, or neither, is ambiguous, so the sink creation fails with this config error.

Source

Thrown at src/connector/src/sink/elasticsearch_opensearch/elasticsearch_opensearch_config.rs:253

            check_username_password()?;
            let transport = transport_builder
                .build()
                .map_err(|e| SinkError::ElasticSearchOpenSearch(anyhow!(e)))?;
            let client = opensearch::OpenSearch::new(transport);
            Ok(ElasticSearchOpenSearchClient::OpenSearch(client))
        } else {
            panic!(
                "connector type must be {} or {}, but get {}",
                ES_SINK, OPENSEARCH_SINK, connector
            );
        }
    }

    pub fn validate_config(&self, schema: &Schema) -> Result<()> {
        if self.index_column.is_some() && self.index.is_some()
            || self.index_column.is_none() && self.index.is_none()
        {
            return Err(SinkError::Config(anyhow!(
                "please set only one of the 'index_column' or 'index' properties."
            )));
        }

        if let Some(index_column) = &self.index_column {
            let filed = schema
                .fields()
                .iter()
                .find(|f| &f.name == index_column)
                .unwrap();
            if filed.data_type() != DataType::Varchar {
                return Err(SinkError::Config(anyhow!(
                    "please ensure the data type of {} is varchar.",
                    index_column
                )));
            }
        }

View on GitHub (pinned to 6469eb736d)

Solutions

1. Keep exactly one of `index` or `index_column` in the WITH clause
2. If you want a fixed target index, set `index` and remove `index_column`
3. If you want per-record dynamic index names, set `index_column` to a varchar column and remove `index`
4. Re-run the CREATE SINK statement after removing the redundant option

Example fix

// before
WITH (connector = 'elasticsearch', index = 'my_index', index_column = 'topic')
// after
WITH (connector = 'elasticsearch', index_column = 'topic')

fn check_index_options(with_opts: &Options) -> Result<(), String> {
    let has_index = with_opts.get("index").is_some();
    let has_index_column = with_opts.get("index_column").is_some();
    if has_index == has_index_column {
        return Err("set exactly one of 'index' or 'index_column'".into());
    }
    Ok(())
}

When it happens

Trigger: CREATE SINK ... WITH (connector='elasticsearch', ...) where (a) both `index` and `index_column` are present in the WITH options, or (b) neither is present. Checked in `validate_config` during sink creation.

Common situations: Copying an example sink definition that already has `index` and adding `index_column` for dynamic indexing; or forgetting to set any index option entirely; or a config template where one of the two was commented out.

Understand the failure class

Background: Conflicting config options: "cannot be used together" — configuration validation errors across open-source libraries — this error's family across 162 libraries.

Related errors

• please ensure the data type of {} is varchar.
• Cannot find {}
• `{}` must be {}, or {}
• Primary key not defined for upsert doris sink (please define
• Can't get fe host from url

AI-assisted analysis of risingwavelabs/risingwave@6469eb736d (2026-09-11). Data as JSON: /api/errors/f4e5e34afbbc14a4. Report an issue: GitHub.