risingwave_connector/sink/
mod.rs

1// Copyright 2022 RisingWave Labs
2//
3// Licensed under the Apache License, Version 2.0 (the "License");
4// you may not use this file except in compliance with the License.
5// You may obtain a copy of the License at
6//
7//     http://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS,
11// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12// See the License for the specific language governing permissions and
13// limitations under the License.
14
15feature_gated_sink_mod!(big_query, "bigquery");
16pub mod boxed;
17pub mod catalog;
18feature_gated_sink_mod!(clickhouse, ClickHouse, "clickhouse");
19pub mod coordinate;
20pub mod decouple_checkpoint_log_sink;
21feature_gated_sink_mod!(deltalake, DeltaLake, "deltalake");
22feature_gated_sink_mod!(doris, "doris");
23#[cfg(any(feature = "sink-doris", feature = "sink-starrocks"))]
24pub mod doris_starrocks_connector;
25pub mod dynamodb;
26pub mod elasticsearch_opensearch;
27pub mod encoder;
28pub mod file_sink;
29pub mod formatter;
30feature_gated_sink_mod!(google_pubsub, GooglePubSub, "google_pubsub");
31pub mod http;
32pub mod iceberg;
33pub mod kafka;
34pub mod kinesis;
35use risingwave_common::bail;
36use risingwave_pb::stream_plan::PbSinkSchemaChange;
37pub mod jdbc_jni_client;
38pub mod log_store;
39pub mod mock_coordination_client;
40pub mod mongodb;
41pub mod mqtt;
42pub mod nats;
43pub mod postgres;
44pub mod pulsar;
45pub mod redis;
46pub mod remote;
47pub mod snowflake_redshift;
48pub mod sqlserver;
49feature_gated_sink_mod!(starrocks, "starrocks");
50pub mod test_sink;
51pub mod trivial;
52pub mod turbopuffer;
53pub mod utils;
54pub mod writer;
55pub mod prelude {
56    pub use crate::sink::{
57        Result, SINK_TYPE_APPEND_ONLY, SINK_USER_FORCE_APPEND_ONLY_OPTION,
58        SINK_USER_FORCE_COMPACTION, Sink, SinkError, SinkParam, SinkWriterParam,
59    };
60}
61
62use std::collections::BTreeMap;
63use std::future::Future;
64use std::sync::{Arc, LazyLock};
65
66use ::redis::RedisError;
67use anyhow::anyhow;
68use async_trait::async_trait;
69use chrono_tz::{Tz, UTC};
70use decouple_checkpoint_log_sink::{
71    COMMIT_CHECKPOINT_INTERVAL, DEFAULT_COMMIT_CHECKPOINT_INTERVAL_WITH_SINK_DECOUPLE,
72    DEFAULT_COMMIT_CHECKPOINT_INTERVAL_WITHOUT_SINK_DECOUPLE,
73};
74use futures::future::BoxFuture;
75use opendal::Error as OpendalError;
76use prometheus::Registry;
77use risingwave_common::array::ArrayError;
78use risingwave_common::bitmap::Bitmap;
79use risingwave_common::catalog::{ColumnDesc, Field, Schema};
80use risingwave_common::config::StreamingConfig;
81use risingwave_common::hash::ActorId;
82use risingwave_common::metrics::{
83    LabelGuardedHistogram, LabelGuardedHistogramVec, LabelGuardedIntCounter,
84    LabelGuardedIntCounterVec, LabelGuardedIntGaugeVec,
85};
86use risingwave_common::monitor::GLOBAL_METRICS_REGISTRY;
87use risingwave_common::secret::{LocalSecretManager, SecretError};
88use risingwave_common::session_config::sink_decouple::SinkDecouple;
89use risingwave_common::{
90    register_guarded_histogram_vec_with_registry, register_guarded_int_counter_vec_with_registry,
91    register_guarded_int_gauge_vec_with_registry,
92};
93use risingwave_pb::catalog::PbSinkType;
94use risingwave_pb::connector_service::{PbSinkParam, SinkMetadata, TableSchema};
95use risingwave_pb::id::ExecutorId;
96use risingwave_rpc_client::MetaClient;
97use risingwave_rpc_client::error::RpcError;
98use starrocks::STARROCKS_SINK;
99use thiserror::Error;
100use thiserror_ext::AsReport;
101use tokio::sync::mpsc::{UnboundedReceiver, UnboundedSender};
102pub use tracing;
103
104use self::catalog::{SinkFormatDesc, SinkType};
105use self::clickhouse::CLICKHOUSE_SINK;
106use self::deltalake::DELTALAKE_SINK;
107use self::iceberg::ICEBERG_SINK;
108use self::mock_coordination_client::{MockMetaClient, SinkCoordinationRpcClientEnum};
109use crate::WithPropertiesExt;
110use crate::connector_common::IcebergSinkCompactionUpdate;
111use crate::error::{ConnectorError, ConnectorResult};
112use crate::sink::boxed::{BoxSinglePhaseCoordinator, BoxTwoPhaseCoordinator};
113use crate::sink::catalog::desc::SinkDesc;
114use crate::sink::catalog::{SinkCatalog, SinkId};
115use crate::sink::decouple_checkpoint_log_sink::ICEBERG_DEFAULT_COMMIT_CHECKPOINT_INTERVAL;
116use crate::sink::file_sink::fs::FsSink;
117use crate::sink::log_store::{LogReader, LogStoreReadItem, LogStoreResult, TruncateOffset};
118use crate::sink::snowflake_redshift::snowflake::SNOWFLAKE_SINK_V2;
119use crate::sink::utils::feature_gated_sink_mod;
120
121const BOUNDED_CHANNEL_SIZE: usize = 16;
122#[macro_export]
123macro_rules! for_all_sinks {
124    ($macro:path $(, $arg:tt)*) => {
125        $macro! {
126            {
127                { Redis, $crate::sink::redis::RedisSink, $crate::sink::redis::RedisConfig },
128                { Kafka, $crate::sink::kafka::KafkaSink, $crate::sink::kafka::KafkaConfig },
129                { Pulsar, $crate::sink::pulsar::PulsarSink, $crate::sink::pulsar::PulsarConfig },
130                { BlackHole, $crate::sink::trivial::BlackHoleSink, () },
131                { Http, $crate::sink::http::HttpSink, $crate::sink::http::HttpConfig },
132                { Turbopuffer, $crate::sink::turbopuffer::TurbopufferSink, $crate::sink::turbopuffer::TurbopufferConfig },
133                { Kinesis, $crate::sink::kinesis::KinesisSink, $crate::sink::kinesis::KinesisSinkConfig },
134                { ClickHouse, $crate::sink::clickhouse::ClickHouseSink, $crate::sink::clickhouse::ClickHouseConfig },
135                { Iceberg, $crate::sink::iceberg::IcebergSink, $crate::sink::iceberg::IcebergConfig },
136                { Mqtt, $crate::sink::mqtt::MqttSink, $crate::sink::mqtt::MqttConfig },
137                { GooglePubSub, $crate::sink::google_pubsub::GooglePubSubSink, $crate::sink::google_pubsub::GooglePubSubConfig },
138                { Nats, $crate::sink::nats::NatsSink, $crate::sink::nats::NatsConfig },
139                { Jdbc, $crate::sink::remote::JdbcSink, () },
140                { ElasticSearch, $crate::sink::elasticsearch_opensearch::elasticsearch::ElasticSearchSink, $crate::sink::elasticsearch_opensearch::elasticsearch_opensearch_config::ElasticSearchConfig },
141                { Opensearch, $crate::sink::elasticsearch_opensearch::opensearch::OpenSearchSink, $crate::sink::elasticsearch_opensearch::elasticsearch_opensearch_config::OpenSearchConfig },
142                { Cassandra, $crate::sink::remote::CassandraSink, () },
143                { Doris, $crate::sink::doris::DorisSink, $crate::sink::doris::DorisConfig },
144                { Starrocks, $crate::sink::starrocks::StarrocksSink, $crate::sink::starrocks::StarrocksConfig },
145                { S3, $crate::sink::file_sink::opendal_sink::FileSink<$crate::sink::file_sink::s3::S3Sink>, $crate::sink::file_sink::s3::S3Config },
146
147                { Gcs, $crate::sink::file_sink::opendal_sink::FileSink<$crate::sink::file_sink::gcs::GcsSink>, $crate::sink::file_sink::gcs::GcsConfig },
148                { Azblob, $crate::sink::file_sink::opendal_sink::FileSink<$crate::sink::file_sink::azblob::AzblobSink>, $crate::sink::file_sink::azblob::AzblobConfig },
149                { Webhdfs, $crate::sink::file_sink::opendal_sink::FileSink<$crate::sink::file_sink::webhdfs::WebhdfsSink>, $crate::sink::file_sink::webhdfs::WebhdfsConfig },
150
151                { Fs, $crate::sink::file_sink::opendal_sink::FileSink<FsSink>, $crate::sink::file_sink::fs::FsConfig },
152                { SnowflakeV2, $crate::sink::snowflake_redshift::snowflake::SnowflakeV2Sink, $crate::sink::snowflake_redshift::snowflake::SnowflakeV2Config },
153                { Snowflake, $crate::sink::file_sink::opendal_sink::FileSink<$crate::sink::file_sink::s3::SnowflakeSink>, $crate::sink::file_sink::s3::SnowflakeConfig },
154                { RedShift, $crate::sink::snowflake_redshift::redshift::RedshiftSink, $crate::sink::snowflake_redshift::redshift::RedShiftConfig },
155                { DeltaLake, $crate::sink::deltalake::DeltaLakeSink, $crate::sink::deltalake::DeltaLakeConfig },
156                { BigQuery, $crate::sink::big_query::BigQuerySink, $crate::sink::big_query::BigQueryConfig },
157                { DynamoDb, $crate::sink::dynamodb::DynamoDbSink, $crate::sink::dynamodb::DynamoDbConfig },
158                { Mongodb, $crate::sink::mongodb::MongodbSink, $crate::sink::mongodb::MongodbConfig },
159                { SqlServer, $crate::sink::sqlserver::SqlServerSink, $crate::sink::sqlserver::SqlServerConfig },
160                { Postgres, $crate::sink::postgres::PostgresSink, $crate::sink::postgres::PostgresConfig },
161
162                { Test, $crate::sink::test_sink::TestSink, () },
163                { Table, $crate::sink::trivial::TableSink, () }
164            }
165            $(,$arg)*
166        }
167    };
168}
169
170#[macro_export]
171macro_rules! generate_config_use_clauses {
172    ({$({ $variant_name:ident, $sink_type:ty, $($config_type:tt)+ }), *}) => {
173        $(
174            $crate::generate_config_use_single! { $($config_type)+ }
175        )*
176    };
177}
178
179#[macro_export]
180macro_rules! generate_config_use_single {
181    // Skip () config types
182    (()) => {};
183
184    // Generate use clause for actual config types
185    ($config_type:path) => {
186        #[allow(unused_imports)]
187        pub(super) use $config_type;
188    };
189}
190
191// Convenience macro that uses for_all_sinks
192#[macro_export]
193macro_rules! use_all_sink_configs {
194    () => {
195        $crate::for_all_sinks! { $crate::generate_config_use_clauses }
196    };
197}
198
199#[macro_export]
200macro_rules! dispatch_sink {
201    ({$({$variant_name:ident, $sink_type:ty, $config_type:ty}),*}, $impl:tt, $sink:tt, $body:tt) => {{
202        use $crate::sink::SinkImpl;
203
204        match $impl {
205            $(
206                SinkImpl::$variant_name($sink) => $body,
207            )*
208        }
209    }};
210    ($impl:expr, $sink:ident, $body:expr) => {{
211        $crate::for_all_sinks! {$crate::dispatch_sink, {$impl}, $sink, {$body}}
212    }};
213}
214
215#[macro_export]
216macro_rules! match_sink_name_str {
217    ({$({$variant_name:ident, $sink_type:ty, $config_type:ty}),*}, $name_str:tt, $type_name:ident, $body:tt, $on_other_closure:tt) => {{
218        use $crate::sink::Sink;
219        match $name_str {
220            $(
221                <$sink_type>::SINK_NAME => {
222                    type $type_name = $sink_type;
223                    {
224                        $body
225                    }
226                },
227            )*
228            other => ($on_other_closure)(other),
229        }
230    }};
231    ($name_str:expr, $type_name:ident, $body:expr, $on_other_closure:expr) => {{
232        $crate::for_all_sinks! {$crate::match_sink_name_str, {$name_str}, $type_name, {$body}, {$on_other_closure}}
233    }};
234}
235
236pub const CONNECTOR_TYPE_KEY: &str = "connector";
237pub const SINK_TYPE_OPTION: &str = "type";
238/// `snapshot = false` corresponds to [`risingwave_pb::stream_plan::StreamScanType::UpstreamOnly`]
239pub const SINK_SNAPSHOT_OPTION: &str = "snapshot";
240pub const SINK_TYPE_APPEND_ONLY: &str = "append-only";
241pub const SINK_TYPE_DEBEZIUM: &str = "debezium";
242pub const SINK_TYPE_UPSERT: &str = "upsert";
243pub const SINK_TYPE_RETRACT: &str = "retract";
244/// Whether to drop DELETE and convert UPDATE to INSERT in the sink executor.
245pub const SINK_USER_IGNORE_DELETE_OPTION: &str = "ignore_delete";
246/// Alias for [`SINK_USER_IGNORE_DELETE_OPTION`], kept for backward compatibility.
247pub const SINK_USER_FORCE_APPEND_ONLY_OPTION: &str = "force_append_only";
248pub const SINK_USER_FORCE_COMPACTION: &str = "force_compaction";
249/// When enabled, the sink executor preserves distinct upstream stream-key changes that map to the
250/// same downstream primary key instead of compacting them into one final-state update within a
251/// barrier. Upstream changes under the same stream key may still be compacted earlier.
252pub const SINK_USER_PRESERVE_ROW_LEVEL_CHANGES: &str = "preserve_row_level_changes";
253
254#[derive(Debug, Clone, PartialEq, Eq)]
255pub struct SinkParam {
256    pub sink_id: SinkId,
257    pub sink_name: String,
258    pub properties: BTreeMap<String, String>,
259    pub columns: Vec<ColumnDesc>,
260    /// User-defined primary key indices for upsert sink, if any.
261    pub downstream_pk: Option<Vec<usize>>,
262    pub sink_type: SinkType,
263    /// Whether to drop DELETE and convert UPDATE to INSERT in the sink executor.
264    pub ignore_delete: bool,
265    pub format_desc: Option<SinkFormatDesc>,
266    pub db_name: String,
267
268    /// - For `CREATE SINK ... FROM ...`, the name of the source table.
269    /// - For `CREATE SINK ... AS <query>`, the name of the sink itself.
270    ///
271    /// See also `gen_sink_plan`.
272    // TODO(eric): Why need these 2 fields (db_name and sink_from_name)?
273    pub sink_from_name: String,
274}
275
276impl SinkParam {
277    pub fn from_proto(pb_param: PbSinkParam) -> Self {
278        let ignore_delete = pb_param.ignore_delete();
279        let table_schema = pb_param.table_schema.expect("should contain table schema");
280        let format_desc = match pb_param.format_desc {
281            Some(f) => f.try_into().ok(),
282            None => {
283                let connector = pb_param.properties.get(CONNECTOR_TYPE_KEY);
284                let r#type = pb_param.properties.get(SINK_TYPE_OPTION);
285                match (connector, r#type) {
286                    (Some(c), Some(t)) => SinkFormatDesc::from_legacy_type(c, t).ok().flatten(),
287                    _ => None,
288                }
289            }
290        };
291        Self {
292            sink_id: SinkId::from(pb_param.sink_id),
293            sink_name: pb_param.sink_name,
294            properties: pb_param.properties,
295            columns: table_schema.columns.iter().map(ColumnDesc::from).collect(),
296            downstream_pk: if table_schema.pk_indices.is_empty() {
297                None
298            } else {
299                Some(
300                    (table_schema.pk_indices.iter())
301                        .map(|i| *i as usize)
302                        .collect(),
303                )
304            },
305            sink_type: SinkType::from_proto(
306                PbSinkType::try_from(pb_param.sink_type).expect("should be able to convert"),
307            ),
308            ignore_delete,
309            format_desc,
310            db_name: pb_param.db_name,
311            sink_from_name: pb_param.sink_from_name,
312        }
313    }
314
315    pub fn to_proto(&self) -> PbSinkParam {
316        PbSinkParam {
317            sink_id: self.sink_id,
318            sink_name: self.sink_name.clone(),
319            properties: self.properties.clone(),
320            table_schema: Some(TableSchema {
321                columns: self.columns.iter().map(|col| col.to_protobuf()).collect(),
322                pk_indices: (self.downstream_pk.as_ref())
323                    .map_or_else(Vec::new, |pk| pk.iter().map(|i| *i as u32).collect()),
324            }),
325            sink_type: self.sink_type.to_proto().into(),
326            format_desc: self.format_desc.as_ref().map(|f| f.to_proto()),
327            db_name: self.db_name.clone(),
328            sink_from_name: self.sink_from_name.clone(),
329            raw_ignore_delete: self.ignore_delete,
330        }
331    }
332
333    pub fn schema(&self) -> Schema {
334        Schema {
335            fields: self.columns.iter().map(Field::from).collect(),
336        }
337    }
338
339    /// Get the downstream primary key indices specified by the user. If not specified, return
340    /// an empty vector.
341    ///
342    /// Prefer directly accessing the `downstream_pk` field, as it uses `None` to represent
343    /// unspecified values, making it clearer.
344    pub fn downstream_pk_or_empty(&self) -> Vec<usize> {
345        self.downstream_pk.clone().unwrap_or_default()
346    }
347
348    // `SinkParams` should only be used when there is a secret context.
349    // FIXME: Use a new type for `SinkFormatDesc` with properties contain filled secrets.
350    pub fn fill_secret_for_format_desc(
351        format_desc: Option<SinkFormatDesc>,
352    ) -> Result<Option<SinkFormatDesc>> {
353        match format_desc {
354            Some(mut format_desc) => {
355                format_desc.options = LocalSecretManager::global()
356                    .fill_secrets(format_desc.options, format_desc.secret_refs.clone())?;
357                Ok(Some(format_desc))
358            }
359            None => Ok(None),
360        }
361    }
362
363    /// Try to convert a `SinkCatalog` to a `SinkParam` and fill the secrets to properties.
364    pub fn try_from_sink_catalog(sink_catalog: SinkCatalog) -> Result<Self> {
365        let columns = sink_catalog
366            .visible_columns()
367            .map(|col| col.column_desc.clone())
368            .collect();
369        let properties_with_secret = LocalSecretManager::global()
370            .fill_secrets(sink_catalog.properties, sink_catalog.secret_refs)?;
371        let format_desc_with_secret = Self::fill_secret_for_format_desc(sink_catalog.format_desc)?;
372        Ok(Self {
373            sink_id: sink_catalog.id,
374            sink_name: sink_catalog.name,
375            properties: properties_with_secret,
376            columns,
377            downstream_pk: sink_catalog.downstream_pk,
378            sink_type: sink_catalog.sink_type,
379            ignore_delete: sink_catalog.ignore_delete,
380            format_desc: format_desc_with_secret,
381            db_name: sink_catalog.db_name,
382            sink_from_name: sink_catalog.sink_from_name,
383        })
384    }
385}
386
387pub fn enforce_secret_sink(props: &impl WithPropertiesExt) -> ConnectorResult<()> {
388    use crate::enforce_secret::EnforceSecret;
389
390    let connector = props
391        .get_connector()
392        .ok_or_else(|| anyhow!("Must specify 'connector' in WITH clause"))?;
393    let key_iter = props.key_iter();
394    match_sink_name_str!(
395        connector.as_str(),
396        PropType,
397        PropType::enforce_secret(key_iter),
398        |other| bail!("connector '{}' is not supported", other)
399    )
400}
401
402pub static GLOBAL_SINK_METRICS: LazyLock<SinkMetrics> =
403    LazyLock::new(|| SinkMetrics::new(&GLOBAL_METRICS_REGISTRY));
404
405#[derive(Clone)]
406pub struct SinkMetrics {
407    pub sink_commit_duration: LabelGuardedHistogramVec,
408    pub connector_sink_rows_received: LabelGuardedIntCounterVec,
409
410    // Log store writer metrics
411    pub log_store_first_write_epoch: LabelGuardedIntGaugeVec,
412    pub log_store_latest_write_epoch: LabelGuardedIntGaugeVec,
413    pub log_store_write_rows: LabelGuardedIntCounterVec,
414
415    // Log store reader metrics
416    pub log_store_latest_read_epoch: LabelGuardedIntGaugeVec,
417    pub log_store_read_rows: LabelGuardedIntCounterVec,
418    pub log_store_read_bytes: LabelGuardedIntCounterVec,
419    pub log_store_reader_wait_new_future_duration_ns: LabelGuardedIntCounterVec,
420
421    // Iceberg metrics
422    pub iceberg_write_qps: LabelGuardedIntCounterVec,
423    pub iceberg_write_latency: LabelGuardedHistogramVec,
424    pub iceberg_rolling_unflushed_data_file: LabelGuardedIntGaugeVec,
425    pub iceberg_position_delete_cache_num: LabelGuardedIntGaugeVec,
426    pub iceberg_partition_num: LabelGuardedIntGaugeVec,
427    pub iceberg_write_bytes: LabelGuardedIntCounterVec,
428    pub iceberg_snapshot_num: LabelGuardedIntGaugeVec,
429}
430
431impl SinkMetrics {
432    pub fn new(registry: &Registry) -> Self {
433        let sink_commit_duration = register_guarded_histogram_vec_with_registry!(
434            "sink_commit_duration",
435            "Duration of commit op in sink",
436            &["actor_id", "connector", "sink_id", "sink_name"],
437            registry
438        )
439        .unwrap();
440
441        let connector_sink_rows_received = register_guarded_int_counter_vec_with_registry!(
442            "connector_sink_rows_received",
443            "Number of rows received by sink",
444            &["actor_id", "connector_type", "sink_id", "sink_name"],
445            registry
446        )
447        .unwrap();
448
449        let log_store_first_write_epoch = register_guarded_int_gauge_vec_with_registry!(
450            "log_store_first_write_epoch",
451            "The first write epoch of log store",
452            &["actor_id", "sink_id", "sink_name"],
453            registry
454        )
455        .unwrap();
456
457        let log_store_latest_write_epoch = register_guarded_int_gauge_vec_with_registry!(
458            "log_store_latest_write_epoch",
459            "The latest write epoch of log store",
460            &["actor_id", "sink_id", "sink_name"],
461            registry
462        )
463        .unwrap();
464
465        let log_store_write_rows = register_guarded_int_counter_vec_with_registry!(
466            "log_store_write_rows",
467            "The write rate of rows",
468            &["actor_id", "sink_id", "sink_name"],
469            registry
470        )
471        .unwrap();
472
473        let log_store_latest_read_epoch = register_guarded_int_gauge_vec_with_registry!(
474            "log_store_latest_read_epoch",
475            "The latest read epoch of log store",
476            &["actor_id", "connector", "sink_id", "sink_name"],
477            registry
478        )
479        .unwrap();
480
481        let log_store_read_rows = register_guarded_int_counter_vec_with_registry!(
482            "log_store_read_rows",
483            "The read rate of rows",
484            &["actor_id", "connector", "sink_id", "sink_name"],
485            registry
486        )
487        .unwrap();
488
489        let log_store_read_bytes = register_guarded_int_counter_vec_with_registry!(
490            "log_store_read_bytes",
491            "Total size of chunks read by log reader",
492            &["actor_id", "connector", "sink_id", "sink_name"],
493            registry
494        )
495        .unwrap();
496
497        let log_store_reader_wait_new_future_duration_ns =
498            register_guarded_int_counter_vec_with_registry!(
499                "log_store_reader_wait_new_future_duration_ns",
500                "Accumulated duration of LogReader to wait for next call to create future",
501                &["actor_id", "connector", "sink_id", "sink_name"],
502                registry
503            )
504            .unwrap();
505
506        let iceberg_write_qps = register_guarded_int_counter_vec_with_registry!(
507            "iceberg_write_qps",
508            "The qps of iceberg writer",
509            &["actor_id", "sink_id", "sink_name"],
510            registry
511        )
512        .unwrap();
513
514        let iceberg_write_latency = register_guarded_histogram_vec_with_registry!(
515            "iceberg_write_latency",
516            "The latency of iceberg writer",
517            &["actor_id", "sink_id", "sink_name"],
518            registry
519        )
520        .unwrap();
521
522        let iceberg_rolling_unflushed_data_file = register_guarded_int_gauge_vec_with_registry!(
523            "iceberg_rolling_unflushed_data_file",
524            "The unflushed data file count of iceberg rolling writer",
525            &["actor_id", "sink_id", "sink_name"],
526            registry
527        )
528        .unwrap();
529
530        let iceberg_position_delete_cache_num = register_guarded_int_gauge_vec_with_registry!(
531            "iceberg_position_delete_cache_num",
532            "The delete cache num of iceberg position delete writer",
533            &["actor_id", "sink_id", "sink_name"],
534            registry
535        )
536        .unwrap();
537
538        let iceberg_partition_num = register_guarded_int_gauge_vec_with_registry!(
539            "iceberg_partition_num",
540            "The partition num of iceberg partition writer",
541            &["actor_id", "sink_id", "sink_name"],
542            registry
543        )
544        .unwrap();
545
546        let iceberg_write_bytes = register_guarded_int_counter_vec_with_registry!(
547            "iceberg_write_bytes",
548            "The write bytes of iceberg writer",
549            &["actor_id", "sink_id", "sink_name"],
550            registry
551        )
552        .unwrap();
553
554        let iceberg_snapshot_num = register_guarded_int_gauge_vec_with_registry!(
555            "iceberg_snapshot_num",
556            "The snapshot number of iceberg table",
557            &["sink_name", "catalog_name", "table_name"],
558            registry
559        )
560        .unwrap();
561
562        Self {
563            sink_commit_duration,
564            connector_sink_rows_received,
565            log_store_first_write_epoch,
566            log_store_latest_write_epoch,
567            log_store_write_rows,
568            log_store_latest_read_epoch,
569            log_store_read_rows,
570            log_store_read_bytes,
571            log_store_reader_wait_new_future_duration_ns,
572            iceberg_write_qps,
573            iceberg_write_latency,
574            iceberg_rolling_unflushed_data_file,
575            iceberg_position_delete_cache_num,
576            iceberg_partition_num,
577            iceberg_write_bytes,
578            iceberg_snapshot_num,
579        }
580    }
581}
582
583#[derive(Clone)]
584pub struct SinkWriterParam {
585    // TODO(eric): deprecate executor_id
586    pub executor_id: ExecutorId,
587    pub vnode_bitmap: Option<Bitmap>,
588    pub meta_client: Option<SinkMetaClient>,
589    // The val has two effect:
590    // 1. Indicates that the sink will accpect the data chunk with extra partition value column.
591    // 2. The index of the extra partition value column.
592    // More detail of partition value column, see `PartitionComputeInfo`
593    pub extra_partition_col_idx: Option<usize>,
594
595    pub actor_id: ActorId,
596    pub sink_id: SinkId,
597    pub sink_name: String,
598    pub connector: String,
599    pub streaming_config: StreamingConfig,
600    pub time_zone: Tz,
601}
602
603#[derive(Clone)]
604pub struct SinkWriterMetrics {
605    pub sink_commit_duration: LabelGuardedHistogram,
606    pub connector_sink_rows_received: LabelGuardedIntCounter,
607}
608
609impl SinkWriterMetrics {
610    pub fn new(writer_param: &SinkWriterParam) -> Self {
611        let labels = [
612            &writer_param.actor_id.to_string(),
613            writer_param.connector.as_str(),
614            &writer_param.sink_id.to_string(),
615            writer_param.sink_name.as_str(),
616        ];
617        let sink_commit_duration = GLOBAL_SINK_METRICS
618            .sink_commit_duration
619            .with_guarded_label_values(&labels);
620        let connector_sink_rows_received = GLOBAL_SINK_METRICS
621            .connector_sink_rows_received
622            .with_guarded_label_values(&labels);
623        Self {
624            sink_commit_duration,
625            connector_sink_rows_received,
626        }
627    }
628
629    #[cfg(test)]
630    pub fn for_test() -> Self {
631        Self {
632            sink_commit_duration: LabelGuardedHistogram::test_histogram::<4>(),
633            connector_sink_rows_received: LabelGuardedIntCounter::test_int_counter::<4>(),
634        }
635    }
636}
637
638#[derive(Clone)]
639pub enum SinkMetaClient {
640    MetaClient(MetaClient),
641    MockMetaClient(MockMetaClient),
642}
643
644impl SinkMetaClient {
645    pub async fn sink_coordinate_client(&self) -> SinkCoordinationRpcClientEnum {
646        match self {
647            SinkMetaClient::MetaClient(meta_client) => {
648                SinkCoordinationRpcClientEnum::SinkCoordinationRpcClient(
649                    meta_client.sink_coordinate_client().await,
650                )
651            }
652            SinkMetaClient::MockMetaClient(mock_meta_client) => {
653                SinkCoordinationRpcClientEnum::MockSinkCoordinationRpcClient(
654                    mock_meta_client.sink_coordinate_client(),
655                )
656            }
657        }
658    }
659
660    pub async fn add_sink_fail_evet_log(
661        &self,
662        sink_id: SinkId,
663        sink_name: String,
664        connector: String,
665        error: String,
666    ) {
667        match self {
668            SinkMetaClient::MetaClient(meta_client) => {
669                match meta_client
670                    .add_sink_fail_evet(sink_id, sink_name, connector, error)
671                    .await
672                {
673                    Ok(_) => {}
674                    Err(e) => {
675                        tracing::warn!(error = %e.as_report(), %sink_id, "Failed to add sink fail event to event log.");
676                    }
677                }
678            }
679            SinkMetaClient::MockMetaClient(_) => {}
680        }
681    }
682}
683
684impl SinkWriterParam {
685    pub fn for_test() -> Self {
686        SinkWriterParam {
687            executor_id: Default::default(),
688            vnode_bitmap: Default::default(),
689            meta_client: Default::default(),
690            extra_partition_col_idx: Default::default(),
691
692            actor_id: 1.into(),
693            sink_id: SinkId::new(1),
694            sink_name: "test_sink".to_owned(),
695            connector: "test_connector".to_owned(),
696            streaming_config: StreamingConfig::default(),
697            time_zone: UTC,
698        }
699    }
700}
701
702fn is_sink_support_commit_checkpoint_interval(sink_name: &str) -> bool {
703    matches!(
704        sink_name,
705        ICEBERG_SINK | CLICKHOUSE_SINK | STARROCKS_SINK | DELTALAKE_SINK | SNOWFLAKE_SINK_V2
706    )
707}
708pub trait Sink: TryFrom<SinkParam, Error = SinkError> {
709    const SINK_NAME: &'static str;
710
711    type LogSinker: LogSinker;
712
713    fn set_default_commit_checkpoint_interval(
714        desc: &mut SinkDesc,
715        user_specified: &SinkDecouple,
716    ) -> Result<()> {
717        if is_sink_support_commit_checkpoint_interval(Self::SINK_NAME) {
718            match desc.properties.get(COMMIT_CHECKPOINT_INTERVAL) {
719                Some(commit_checkpoint_interval) => {
720                    let commit_checkpoint_interval = commit_checkpoint_interval
721                        .parse::<u64>()
722                        .map_err(|e| SinkError::Config(anyhow!(e)))?;
723                    if matches!(user_specified, SinkDecouple::Disable)
724                        && commit_checkpoint_interval > 1
725                    {
726                        return Err(SinkError::Config(anyhow!(
727                            "config conflict: `commit_checkpoint_interval` larger than 1 means that sink decouple must be enabled, but session config sink_decouple is disabled"
728                        )));
729                    }
730                }
731                None => match user_specified {
732                    SinkDecouple::Default | SinkDecouple::Enable => {
733                        if matches!(Self::SINK_NAME, ICEBERG_SINK) {
734                            desc.properties.insert(
735                                COMMIT_CHECKPOINT_INTERVAL.to_owned(),
736                                ICEBERG_DEFAULT_COMMIT_CHECKPOINT_INTERVAL.to_string(),
737                            );
738                        } else {
739                            desc.properties.insert(
740                                COMMIT_CHECKPOINT_INTERVAL.to_owned(),
741                                DEFAULT_COMMIT_CHECKPOINT_INTERVAL_WITH_SINK_DECOUPLE.to_string(),
742                            );
743                        }
744                    }
745                    SinkDecouple::Disable => {
746                        desc.properties.insert(
747                            COMMIT_CHECKPOINT_INTERVAL.to_owned(),
748                            DEFAULT_COMMIT_CHECKPOINT_INTERVAL_WITHOUT_SINK_DECOUPLE.to_string(),
749                        );
750                    }
751                },
752            }
753        }
754        Ok(())
755    }
756
757    /// `user_specified` is the value of `sink_decouple` config.
758    fn is_sink_decouple(user_specified: &SinkDecouple) -> Result<bool> {
759        match user_specified {
760            SinkDecouple::Default | SinkDecouple::Enable => Ok(true),
761            SinkDecouple::Disable => Ok(false),
762        }
763    }
764
765    fn support_schema_change() -> bool {
766        false
767    }
768
769    fn validate_alter_config(_config: &BTreeMap<String, String>) -> Result<()> {
770        Ok(())
771    }
772
773    async fn validate(&self) -> Result<()>;
774    async fn new_log_sinker(&self, writer_param: SinkWriterParam) -> Result<Self::LogSinker>;
775
776    fn is_coordinated_sink(&self) -> bool {
777        false
778    }
779
780    async fn new_coordinator(
781        &self,
782        _iceberg_compact_stat_sender: Option<UnboundedSender<IcebergSinkCompactionUpdate>>,
783    ) -> Result<SinkCommitCoordinator> {
784        Err(SinkError::Coordinator(anyhow!("no coordinator")))
785    }
786}
787
788pub trait SinkLogReader: Send {
789    fn start_from(
790        &mut self,
791        start_offset: Option<u64>,
792    ) -> impl Future<Output = LogStoreResult<()>> + Send + '_;
793    /// Emit the next item.
794    ///
795    /// The implementation should ensure that the future is cancellation safe.
796    fn next_item(
797        &mut self,
798    ) -> impl Future<Output = LogStoreResult<(u64, LogStoreReadItem)>> + Send + '_;
799
800    /// Mark that all items emitted so far have been consumed and it is safe to truncate the log
801    /// from the current offset.
802    fn truncate(&mut self, offset: TruncateOffset) -> LogStoreResult<()>;
803}
804
805impl<R: LogReader> SinkLogReader for &mut R {
806    fn next_item(
807        &mut self,
808    ) -> impl Future<Output = LogStoreResult<(u64, LogStoreReadItem)>> + Send + '_ {
809        <R as LogReader>::next_item(*self)
810    }
811
812    fn truncate(&mut self, offset: TruncateOffset) -> LogStoreResult<()> {
813        <R as LogReader>::truncate(*self, offset)
814    }
815
816    fn start_from(
817        &mut self,
818        start_offset: Option<u64>,
819    ) -> impl Future<Output = LogStoreResult<()>> + Send + '_ {
820        <R as LogReader>::start_from(*self, start_offset)
821    }
822}
823
824#[async_trait]
825pub trait LogSinker: 'static + Send {
826    // Note: Please rebuild the log reader's read stream before consuming the log store,
827    async fn consume_log_and_sink(self, log_reader: impl SinkLogReader) -> Result<!>;
828}
829pub type SinkCommittedEpochSubscriber = Arc<
830    dyn Fn(SinkId) -> BoxFuture<'static, Result<(u64, UnboundedReceiver<u64>)>>
831        + Send
832        + Sync
833        + 'static,
834>;
835
836pub enum SinkCommitCoordinator {
837    SinglePhase(BoxSinglePhaseCoordinator),
838    TwoPhase(BoxTwoPhaseCoordinator),
839}
840
841#[async_trait]
842pub trait SinglePhaseCommitCoordinator {
843    /// Initialize the sink committer coordinator.
844    async fn init(&mut self) -> Result<()>;
845
846    /// Commit data directly using single-phase strategy.
847    async fn commit_data(&mut self, epoch: u64, metadata: Vec<SinkMetadata>) -> Result<()>;
848
849    /// Idempotent implementation is required, because `commit_schema_change` in the same epoch could be called multiple
850    /// times.
851    async fn commit_schema_change(
852        &mut self,
853        _epoch: u64,
854        _schema_change: PbSinkSchemaChange,
855    ) -> Result<()> {
856        Err(SinkError::Coordinator(anyhow!(
857            "Schema change is not implemented for single-phase commit coordinator {}",
858            std::any::type_name::<Self>()
859        )))
860    }
861}
862
863#[async_trait]
864pub trait TwoPhaseCommitCoordinator {
865    /// Initialize the sink committer coordinator.
866    async fn init(&mut self) -> Result<()>;
867
868    /// Return serialized commit metadata to be passed to `commit`.
869    async fn pre_commit(
870        &mut self,
871        epoch: u64,
872        metadata: Vec<SinkMetadata>,
873        schema_change: Option<PbSinkSchemaChange>,
874    ) -> Result<Option<Vec<u8>>>;
875
876    /// Idempotent implementation is required, because `commit_data` in the same epoch could be called multiple times.
877    async fn commit_data(&mut self, epoch: u64, commit_metadata: Vec<u8>) -> Result<()>;
878
879    /// Idempotent implementation is required, because `commit_schema_change` in the same epoch could be called multiple
880    /// times.
881    async fn commit_schema_change(
882        &mut self,
883        _epoch: u64,
884        _schema_change: PbSinkSchemaChange,
885    ) -> Result<()> {
886        Err(SinkError::Coordinator(anyhow!(
887            "Schema change is not implemented for two-phase commit coordinator {}",
888            std::any::type_name::<Self>()
889        )))
890    }
891
892    /// Idempotent implementation is required, because `abort` in the same epoch could be called multiple times.
893    async fn abort(&mut self, epoch: u64, commit_metadata: Vec<u8>);
894}
895
896impl SinkImpl {
897    pub fn new(mut param: SinkParam) -> Result<Self> {
898        const PRIVATE_LINK_TARGET_KEY: &str = "privatelink.targets";
899
900        // remove privatelink related properties if any
901        param.properties.remove(PRIVATE_LINK_TARGET_KEY);
902
903        let sink_type = param
904            .properties
905            .get(CONNECTOR_TYPE_KEY)
906            .ok_or_else(|| SinkError::Config(anyhow!("missing config: {}", CONNECTOR_TYPE_KEY)))?;
907
908        let sink_type = sink_type.to_lowercase();
909        match_sink_name_str!(
910            sink_type.as_str(),
911            SinkType,
912            Ok(SinkType::try_from(param)?.into()),
913            |other| {
914                Err(SinkError::Config(anyhow!(
915                    "unsupported sink connector {}",
916                    other
917                )))
918            }
919        )
920    }
921
922    pub fn is_sink_into_table(&self) -> bool {
923        matches!(self, SinkImpl::Table(_))
924    }
925
926    pub fn is_blackhole(&self) -> bool {
927        matches!(self, SinkImpl::BlackHole(_))
928    }
929
930    pub fn is_coordinated_sink(&self) -> bool {
931        dispatch_sink!(self, sink, sink.is_coordinated_sink())
932    }
933}
934
935pub fn build_sink(param: SinkParam) -> Result<SinkImpl> {
936    SinkImpl::new(param)
937}
938
939macro_rules! def_sink_impl {
940    () => {
941        $crate::for_all_sinks! { def_sink_impl }
942    };
943    ({ $({ $variant_name:ident, $sink_type:ty, $config_type:ty }),* }) => {
944        #[derive(Debug)]
945        pub enum SinkImpl {
946            $(
947                $variant_name(Box<$sink_type>),
948            )*
949        }
950
951        $(
952            impl From<$sink_type> for SinkImpl {
953                fn from(sink: $sink_type) -> SinkImpl {
954                    SinkImpl::$variant_name(Box::new(sink))
955                }
956            }
957        )*
958    };
959}
960
961def_sink_impl!();
962
963pub type Result<T> = std::result::Result<T, SinkError>;
964
965#[derive(Error, Debug)]
966pub enum SinkError {
967    #[error("Kafka error: {0}")]
968    Kafka(#[from] rdkafka::error::KafkaError),
969    #[error("Kinesis error: {0}")]
970    Kinesis(
971        #[source]
972        #[backtrace]
973        anyhow::Error,
974    ),
975    #[error("Remote sink error: {0}")]
976    Remote(
977        #[source]
978        #[backtrace]
979        anyhow::Error,
980    ),
981    #[error("Encode error: {0}")]
982    Encode(String),
983    #[error("Avro error: {0}")]
984    Avro(#[from] apache_avro::Error),
985    #[error("Iceberg error: {0}")]
986    Iceberg(
987        #[source]
988        #[backtrace]
989        anyhow::Error,
990    ),
991    #[error("config error: {0}")]
992    Config(
993        #[source]
994        #[backtrace]
995        anyhow::Error,
996    ),
997    #[error("coordinator error: {0}")]
998    Coordinator(
999        #[source]
1000        #[backtrace]
1001        anyhow::Error,
1002    ),
1003    #[error("ClickHouse error: {0}")]
1004    ClickHouse(String),
1005    #[error("Redis error: {0}")]
1006    Redis(String),
1007    #[error("Http error: {0}")]
1008    Http(
1009        #[source]
1010        #[backtrace]
1011        anyhow::Error,
1012    ),
1013    #[error("Mqtt error: {0}")]
1014    Mqtt(
1015        #[source]
1016        #[backtrace]
1017        anyhow::Error,
1018    ),
1019    #[error("Nats error: {0}")]
1020    Nats(
1021        #[source]
1022        #[backtrace]
1023        anyhow::Error,
1024    ),
1025    #[error("Google Pub/Sub error: {0}")]
1026    GooglePubSub(
1027        #[source]
1028        #[backtrace]
1029        anyhow::Error,
1030    ),
1031    #[error("Doris/Starrocks connect error: {0}")]
1032    DorisStarrocksConnect(
1033        #[source]
1034        #[backtrace]
1035        anyhow::Error,
1036    ),
1037    #[error("Doris error: {0}")]
1038    Doris(String),
1039    #[error("DeltaLake error: {0}")]
1040    DeltaLake(
1041        #[source]
1042        #[backtrace]
1043        anyhow::Error,
1044    ),
1045    #[error("ElasticSearch/OpenSearch error: {0}")]
1046    ElasticSearchOpenSearch(
1047        #[source]
1048        #[backtrace]
1049        anyhow::Error,
1050    ),
1051    #[error("Starrocks error: {0}")]
1052    Starrocks(String),
1053    #[error("File error: {0}")]
1054    File(String),
1055    #[error("Pulsar error: {0}")]
1056    Pulsar(
1057        #[source]
1058        #[backtrace]
1059        anyhow::Error,
1060    ),
1061    #[error(transparent)]
1062    Internal(
1063        #[from]
1064        #[backtrace]
1065        anyhow::Error,
1066    ),
1067    #[error("BigQuery error: {0}")]
1068    BigQuery(
1069        #[source]
1070        #[backtrace]
1071        anyhow::Error,
1072    ),
1073    #[error("DynamoDB error: {0}")]
1074    DynamoDb(
1075        #[source]
1076        #[backtrace]
1077        anyhow::Error,
1078    ),
1079    #[error("SQL Server error: {0}")]
1080    SqlServer(
1081        #[source]
1082        #[backtrace]
1083        anyhow::Error,
1084    ),
1085    #[error("Postgres error: {0}")]
1086    Postgres(
1087        #[source]
1088        #[backtrace]
1089        anyhow::Error,
1090    ),
1091    #[error(transparent)]
1092    Connector(
1093        #[from]
1094        #[backtrace]
1095        ConnectorError,
1096    ),
1097    #[error("Secret error: {0}")]
1098    Secret(
1099        #[from]
1100        #[backtrace]
1101        SecretError,
1102    ),
1103    #[error("Mongodb error: {0}")]
1104    Mongodb(
1105        #[source]
1106        #[backtrace]
1107        anyhow::Error,
1108    ),
1109    #[error("Redshift error: {0}")]
1110    Redshift(
1111        #[source]
1112        #[backtrace]
1113        anyhow::Error,
1114    ),
1115}
1116
1117#[allow(clippy::disallowed_types)]
1118impl From<::iceberg::Error> for SinkError {
1119    fn from(err: ::iceberg::Error) -> Self {
1120        SinkError::Iceberg(anyhow!(err))
1121    }
1122}
1123
1124impl From<sea_orm::DbErr> for SinkError {
1125    fn from(err: sea_orm::DbErr) -> Self {
1126        SinkError::Iceberg(anyhow!(err))
1127    }
1128}
1129
1130impl From<OpendalError> for SinkError {
1131    fn from(error: OpendalError) -> Self {
1132        SinkError::File(error.to_report_string())
1133    }
1134}
1135
1136impl From<parquet::errors::ParquetError> for SinkError {
1137    fn from(error: parquet::errors::ParquetError) -> Self {
1138        SinkError::File(error.to_report_string())
1139    }
1140}
1141
1142impl From<ArrayError> for SinkError {
1143    fn from(error: ArrayError) -> Self {
1144        SinkError::File(error.to_report_string())
1145    }
1146}
1147
1148impl From<RpcError> for SinkError {
1149    fn from(value: RpcError) -> Self {
1150        SinkError::Remote(anyhow!(value))
1151    }
1152}
1153
1154impl From<RedisError> for SinkError {
1155    fn from(value: RedisError) -> Self {
1156        SinkError::Redis(value.to_report_string())
1157    }
1158}
1159
1160impl From<tiberius::error::Error> for SinkError {
1161    fn from(err: tiberius::error::Error) -> Self {
1162        SinkError::SqlServer(anyhow!(err))
1163    }
1164}
1165
1166impl From<::elasticsearch::Error> for SinkError {
1167    fn from(err: ::elasticsearch::Error) -> Self {
1168        SinkError::ElasticSearchOpenSearch(anyhow!(err))
1169    }
1170}
1171
1172impl From<::opensearch::Error> for SinkError {
1173    fn from(err: ::opensearch::Error) -> Self {
1174        SinkError::ElasticSearchOpenSearch(anyhow!(err))
1175    }
1176}
1177
1178impl From<tokio_postgres::Error> for SinkError {
1179    fn from(err: tokio_postgres::Error) -> Self {
1180        SinkError::Postgres(anyhow!(err))
1181    }
1182}
risingwave_connector/sink/mod.rs

risingwave_connector/sink/
mod.rs
