risingwave_connector/sink/file_sink/
batching_log_sink.rs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
// Copyright 2024 RisingWave Labs
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

use async_trait::async_trait;

use crate::sink::file_sink::opendal_sink::OpenDalSinkWriter;
use crate::sink::log_store::{LogStoreReadItem, TruncateOffset};
use crate::sink::{LogSinker, Result, SinkLogReader};

/// `BatchingLogSinker` is used for a commit-decoupled sink that supports cross-barrier batching.
/// Currently, it is only used for file sinks, so it contains an `OpenDalSinkWriter`.
pub struct BatchingLogSinker {
    writer: OpenDalSinkWriter,
}

impl BatchingLogSinker {
    /// Create a log sinker with a file sink writer.
    pub fn new(writer: OpenDalSinkWriter) -> Self {
        BatchingLogSinker { writer }
    }
}

#[async_trait]
impl LogSinker for BatchingLogSinker {
    async fn consume_log_and_sink(self, log_reader: &mut impl SinkLogReader) -> Result<!> {
        let mut sink_writer = self.writer;
        #[derive(Debug)]
        enum LogConsumerState {
            /// Mark that the log consumer is not initialized yet
            Uninitialized,

            /// Mark that a new epoch has begun, and store the max_uncommitted_epoch for cross-barrier batching.
            /// For example, suppose the current order is (chunk1, barrier1, chunk2, barrier2, chunk3), and the batching is not completed until chunk3,
            /// that is, barrier2 and its previous chunks are not truncated, the `max_uncommitted_epoch` is barrier2.
            /// When we truncate chunk3, we should first truncate barrier2, and then truncate chunk3.
            EpochBegun {
                curr_epoch: u64,
                max_uncommitted_epoch: Option<u64>,
            },

            /// Mark that the consumer has just received a barrier
            BarrierReceived { prev_epoch: u64 },
        }

        let mut state = LogConsumerState::Uninitialized;
        loop {
            let (epoch, item): (u64, LogStoreReadItem) = log_reader.next_item().await?;
            if let LogStoreReadItem::UpdateVnodeBitmap(_) = &item {
                match &state {
                    LogConsumerState::BarrierReceived { prev_epoch } => {
                        // we need to force to finish the batch here. Otherwise, there can be data loss because actor can be dropped and rebuilt during scaling.
                        if sink_writer.try_commit().await? {
                            // If epoch increased, we first need to truncate the previous epoch.
                            if epoch > *prev_epoch {
                                log_reader
                                    .truncate(TruncateOffset::Barrier { epoch: *prev_epoch })?;
                            }
                        };
                    }
                    _ => unreachable!(
                        "update vnode bitmap can be accepted only right after \
                    barrier, but current state is {:?}",
                        state
                    ),
                }
                continue;
            }
            // begin_epoch when not previously began
            state = match state {
                LogConsumerState::Uninitialized => LogConsumerState::EpochBegun {
                    curr_epoch: epoch,
                    max_uncommitted_epoch: None,
                },
                LogConsumerState::EpochBegun {
                    curr_epoch,
                    max_uncommitted_epoch,
                } => {
                    assert!(
                        epoch >= curr_epoch,
                        "new epoch {} should not be below the current epoch {}",
                        epoch,
                        curr_epoch
                    );
                    LogConsumerState::EpochBegun {
                        curr_epoch: epoch,
                        max_uncommitted_epoch,
                    }
                }
                LogConsumerState::BarrierReceived { prev_epoch } => {
                    assert!(
                        epoch > prev_epoch,
                        "new epoch {} should be greater than prev epoch {}",
                        epoch,
                        prev_epoch
                    );
                    LogConsumerState::EpochBegun {
                        curr_epoch: epoch,
                        max_uncommitted_epoch: Some(prev_epoch),
                    }
                }
            };
            match item {
                LogStoreReadItem::StreamChunk { chunk, chunk_id } => {
                    sink_writer.write_batch(chunk).await?;
                    match sink_writer.try_commit().await {
                        Err(e) => {
                            return Err(e);
                        }
                        // The file has been successfully written and is now visible to downstream consumers.
                        // Truncate the file to remove the specified `chunk_id` and any preceding content.
                        Ok(true) => {
                            // If epoch increased, we first need to truncate the previous epoch.
                            if let Some(max_uncommitted_epoch) = match state {
                                LogConsumerState::EpochBegun {
                                    curr_epoch: _,
                                    max_uncommitted_epoch,
                                } => max_uncommitted_epoch,
                                _ => unreachable!("epoch must have begun before handling barrier"),
                            } {
                                assert!(epoch > max_uncommitted_epoch);
                                log_reader.truncate(TruncateOffset::Barrier {
                                    epoch: max_uncommitted_epoch,
                                })?;
                                state = LogConsumerState::EpochBegun {
                                    curr_epoch: epoch,
                                    max_uncommitted_epoch: None,
                                }
                            };

                            log_reader.truncate(TruncateOffset::Chunk {
                                epoch: (epoch),
                                chunk_id: (chunk_id),
                            })?;
                        }
                        // The file has not been written into downstream file system.
                        Ok(false) => {}
                    }
                }
                LogStoreReadItem::Barrier { is_checkpoint: _ } => {
                    let prev_epoch = match state {
                        LogConsumerState::EpochBegun {
                            curr_epoch,
                            max_uncommitted_epoch: _,
                        } => curr_epoch,
                        _ => unreachable!("epoch must have begun before handling barrier"),
                    };

                    // When the barrier arrives, call the writer's try_finish interface to check if the file write can be completed.
                    // If it is completed, which means the file is visible in the downstream file system, then truncate the file in the log store; otherwise, do nothing.
                    // Since the current data must be before the current epoch, we only need to truncate `prev_epoch`.
                    if sink_writer.try_commit().await? {
                        log_reader.truncate(TruncateOffset::Barrier { epoch: prev_epoch })?;
                    };

                    state = LogConsumerState::BarrierReceived { prev_epoch }
                }
                LogStoreReadItem::UpdateVnodeBitmap(_vnode_bitmap) => {}
            }
        }
    }
}