risingwave_meta/manager/iceberg_compaction/
mod.rs1mod gc;
16mod manual;
17mod schedule;
18mod stream;
19
20use std::collections::{BTreeMap, HashMap, HashSet};
21use std::sync::Arc;
22use std::time::Duration;
23
24use anyhow::anyhow;
25use parking_lot::RwLock;
26use risingwave_common::id::WorkerId;
27use risingwave_connector::sink::SinkParam;
28use risingwave_connector::sink::catalog::{SinkCatalog, SinkId};
29use risingwave_connector::sink::iceberg::{ICEBERG_SINK, IcebergConfig};
30use risingwave_connector::source::UPSTREAM_SOURCE_KEY;
31use risingwave_pb::iceberg_compaction::SubscribeIcebergCompactionEventRequest;
32use risingwave_pb::id::IcebergCompactionTaskId;
33use tokio::sync::mpsc::{UnboundedReceiver, UnboundedSender};
34use tonic::Streaming;
35
36use super::MetaSrvEnv;
37use crate::MetaResult;
38use crate::controller::streaming_job::AbortCreatingJobResult;
39use crate::hummock::IcebergCompactorManagerRef;
40use crate::manager::MetadataManager;
41use crate::rpc::metrics::MetaMetrics;
42
43pub type IcebergCompactionManagerRef = Arc<IcebergCompactionManager>;
44
45pub(crate) type CompactorChangeTx =
46 UnboundedSender<(WorkerId, Streaming<SubscribeIcebergCompactionEventRequest>)>;
47
48pub(crate) type CompactorChangeRx =
49 UnboundedReceiver<(WorkerId, Streaming<SubscribeIcebergCompactionEventRequest>)>;
50
51type ManualCompactionWaiter = tokio::sync::oneshot::Sender<MetaResult<IcebergCompactionTaskId>>;
52
53use schedule::CompactionTrack;
54pub use schedule::IcebergCompactionScheduleStatus;
55
56pub struct IcebergCompactionManager {
57 pub env: MetaSrvEnv,
58 inner: Arc<RwLock<IcebergCompactionManagerInner>>,
59
60 metadata_manager: MetadataManager,
61 pub iceberg_compactor_manager: IcebergCompactorManagerRef,
62
63 compactor_streams_change_tx: CompactorChangeTx,
64
65 pub metrics: Arc<MetaMetrics>,
66}
67
68struct IcebergCompactionManagerInner {
69 sink_schedules: HashMap<SinkId, CompactionTrack>,
70 snapshot_expiration_sink_ids: HashSet<SinkId>,
71 manifest_rewrite_sink_ids: HashSet<SinkId>,
72 manual_compaction_waiters: HashMap<SinkId, ManualCompactionWaiter>,
73}
74
75impl IcebergCompactionManager {
76 fn report_timeout(&self) -> Duration {
77 Duration::from_secs(self.env.opts.iceberg_compaction_report_timeout_sec)
78 }
79
80 fn config_refresh_interval(&self) -> Duration {
81 Duration::from_secs(self.env.opts.iceberg_compaction_config_refresh_interval_sec)
82 }
83
84 pub fn build(
85 env: MetaSrvEnv,
86 metadata_manager: MetadataManager,
87 iceberg_compactor_manager: IcebergCompactorManagerRef,
88 metrics: Arc<MetaMetrics>,
89 ) -> (Arc<Self>, CompactorChangeRx) {
90 let (compactor_streams_change_tx, compactor_streams_change_rx) =
91 tokio::sync::mpsc::unbounded_channel();
92 (
93 Arc::new(Self {
94 env,
95 inner: Arc::new(RwLock::new(IcebergCompactionManagerInner {
96 sink_schedules: HashMap::default(),
97 snapshot_expiration_sink_ids: HashSet::default(),
98 manifest_rewrite_sink_ids: HashSet::default(),
99 manual_compaction_waiters: HashMap::default(),
100 })),
101 metadata_manager,
102 iceberg_compactor_manager,
103 compactor_streams_change_tx,
104 metrics,
105 }),
106 compactor_streams_change_rx,
107 )
108 }
109
110 async fn get_sink_param(&self, sink_id: SinkId) -> MetaResult<SinkParam> {
111 let prost_sink_catalog = self
112 .metadata_manager
113 .catalog_controller
114 .get_sink_by_id(sink_id)
115 .await?
116 .ok_or_else(|| anyhow!("Sink not found: {}", sink_id))?;
117 let sink_catalog = SinkCatalog::from(prost_sink_catalog);
118 let param = SinkParam::try_from_sink_catalog(sink_catalog)?;
119 Ok(param)
120 }
121
122 async fn load_iceberg_config(&self, sink_id: SinkId) -> MetaResult<IcebergConfig> {
123 let sink_param = self.get_sink_param(sink_id).await?;
124 let iceberg_config = IcebergConfig::from_btreemap(sink_param.properties)?;
125 Ok(iceberg_config)
126 }
127
128 pub fn clear_maintenance_for_aborted_job(&self, abort_result: &AbortCreatingJobResult) {
131 if let Some(sink_id) = abort_result.aborted_sink_id {
132 self.clear_iceberg_maintenance_by_sink_id(sink_id);
133 }
134 }
135}
136
137pub fn is_iceberg_sink(properties: &BTreeMap<String, String>) -> bool {
140 properties
141 .get(UPSTREAM_SOURCE_KEY)
142 .is_some_and(|connector| connector.eq_ignore_ascii_case(ICEBERG_SINK))
143}