mito2/cache/
write_cache.rs

1// Copyright 2023 Greptime Team
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
15//! A write-through cache for remote object stores.
16
17use std::sync::Arc;
18use std::time::{Duration, Instant};
19
20use common_base::readable_size::ReadableSize;
21use common_telemetry::{debug, info};
22use futures::AsyncWriteExt;
23use object_store::ObjectStore;
24use snafu::ResultExt;
25use store_api::storage::RegionId;
26use tokio::sync::mpsc::{UnboundedSender, unbounded_channel};
27
28use crate::access_layer::{
29    FilePathProvider, Metrics, RegionFilePathFactory, SstInfoArray, SstWriteRequest,
30    TempFileCleaner, WriteCachePathProvider, WriteType, new_fs_cache_store,
31};
32use crate::cache::file_cache::{FileCache, FileCacheRef, FileType, IndexKey, IndexValue};
33use crate::cache::manifest_cache::ManifestCache;
34use crate::error::{self, Result};
35use crate::metrics::UPLOAD_BYTES_TOTAL;
36use crate::region::opener::RegionLoadCacheTask;
37use crate::sst::file::RegionFileId;
38use crate::sst::index::IndexerBuilderImpl;
39use crate::sst::index::intermediate::IntermediateManager;
40use crate::sst::index::puffin_manager::{PuffinManagerFactory, SstPuffinManager};
41use crate::sst::parquet::writer::ParquetWriter;
42use crate::sst::parquet::{SstInfo, WriteOptions};
43use crate::sst::{DEFAULT_WRITE_BUFFER_SIZE, DEFAULT_WRITE_CONCURRENCY};
44
45/// A cache for uploading files to remote object stores.
46///
47/// It keeps files in local disk and then sends files to object stores.
48pub struct WriteCache {
49    /// Local file cache.
50    file_cache: FileCacheRef,
51    /// Puffin manager factory for index.
52    puffin_manager_factory: PuffinManagerFactory,
53    /// Intermediate manager for index.
54    intermediate_manager: IntermediateManager,
55    /// Sender for region load cache tasks.
56    task_sender: UnboundedSender<RegionLoadCacheTask>,
57    /// Optional cache for manifest files.
58    manifest_cache: Option<ManifestCache>,
59}
60
61pub type WriteCacheRef = Arc<WriteCache>;
62
63impl WriteCache {
64    /// Create the cache with a `local_store` to cache files and a
65    /// `object_store_manager` for all object stores.
66    #[allow(clippy::too_many_arguments)]
67    pub async fn new(
68        local_store: ObjectStore,
69        cache_capacity: ReadableSize,
70        ttl: Option<Duration>,
71        index_cache_percent: Option<u8>,
72        enable_background_worker: bool,
73        puffin_manager_factory: PuffinManagerFactory,
74        intermediate_manager: IntermediateManager,
75        manifest_cache: Option<ManifestCache>,
76    ) -> Result<Self> {
77        let (task_sender, task_receiver) = unbounded_channel();
78
79        let file_cache = Arc::new(FileCache::new(
80            local_store,
81            cache_capacity,
82            ttl,
83            index_cache_percent,
84            enable_background_worker,
85        ));
86        file_cache.recover(false, Some(task_receiver)).await;
87
88        Ok(Self {
89            file_cache,
90            puffin_manager_factory,
91            intermediate_manager,
92            task_sender,
93            manifest_cache,
94        })
95    }
96
97    /// Creates a write cache based on local fs.
98    #[allow(clippy::too_many_arguments)]
99    pub async fn new_fs(
100        cache_dir: &str,
101        cache_capacity: ReadableSize,
102        ttl: Option<Duration>,
103        index_cache_percent: Option<u8>,
104        enable_background_worker: bool,
105        puffin_manager_factory: PuffinManagerFactory,
106        intermediate_manager: IntermediateManager,
107        manifest_cache_capacity: ReadableSize,
108    ) -> Result<Self> {
109        info!("Init write cache on {cache_dir}, capacity: {cache_capacity}");
110
111        let local_store = new_fs_cache_store(cache_dir).await?;
112
113        // Create manifest cache if capacity is non-zero
114        let manifest_cache = if manifest_cache_capacity.as_bytes() > 0 {
115            Some(ManifestCache::new(local_store.clone(), manifest_cache_capacity, ttl, false).await)
116        } else {
117            None
118        };
119
120        Self::new(
121            local_store,
122            cache_capacity,
123            ttl,
124            index_cache_percent,
125            enable_background_worker,
126            puffin_manager_factory,
127            intermediate_manager,
128            manifest_cache,
129        )
130        .await
131    }
132
133    /// Returns the file cache of the write cache.
134    pub(crate) fn file_cache(&self) -> FileCacheRef {
135        self.file_cache.clone()
136    }
137
138    /// Returns the manifest cache if available.
139    pub(crate) fn manifest_cache(&self) -> Option<ManifestCache> {
140        self.manifest_cache.clone()
141    }
142
143    /// Build the puffin manager
144    pub(crate) fn build_puffin_manager(&self) -> SstPuffinManager {
145        let store = self.file_cache.local_store();
146        let path_provider = WriteCachePathProvider::new(self.file_cache.clone());
147        self.puffin_manager_factory.build(store, path_provider)
148    }
149
150    /// Put encoded SST data to the cache and upload to the remote object store.
151    pub(crate) async fn put_and_upload_sst(
152        &self,
153        data: &bytes::Bytes,
154        region_id: RegionId,
155        sst_info: &SstInfo,
156        upload_request: SstUploadRequest,
157    ) -> Result<Metrics> {
158        let file_id = sst_info.file_id;
159        let mut metrics = Metrics::new(WriteType::Flush);
160
161        // Create index key for the SST file
162        let parquet_key = IndexKey::new(region_id, file_id, FileType::Parquet);
163
164        // Write to cache first
165        let cache_start = Instant::now();
166        let cache_path = self.file_cache.cache_file_path(parquet_key);
167        let store = self.file_cache.local_store();
168        let cleaner = TempFileCleaner::new(region_id, store.clone());
169        let write_res = store
170            .write(&cache_path, data.clone())
171            .await
172            .context(crate::error::OpenDalSnafu);
173        if let Err(e) = write_res {
174            cleaner.clean_by_file_id(file_id).await;
175            return Err(e);
176        }
177
178        metrics.write_batch = cache_start.elapsed();
179
180        // Upload to remote store
181        let upload_start = Instant::now();
182        let region_file_id = RegionFileId::new(region_id, file_id);
183        let remote_path = upload_request
184            .dest_path_provider
185            .build_sst_file_path(region_file_id);
186
187        if let Err(e) = self
188            .upload(parquet_key, &remote_path, &upload_request.remote_store)
189            .await
190        {
191            // Clean up cache on failure
192            self.remove(parquet_key).await;
193            return Err(e);
194        }
195
196        metrics.upload_parquet = upload_start.elapsed();
197        Ok(metrics)
198    }
199
200    /// Returns the intermediate manager of the write cache.
201    pub(crate) fn intermediate_manager(&self) -> &IntermediateManager {
202        &self.intermediate_manager
203    }
204
205    /// Writes SST to the cache and then uploads it to the remote object store.
206    pub(crate) async fn write_and_upload_sst(
207        &self,
208        write_request: SstWriteRequest,
209        upload_request: SstUploadRequest,
210        write_opts: &WriteOptions,
211        metrics: &mut Metrics,
212    ) -> Result<SstInfoArray> {
213        let region_id = write_request.metadata.region_id;
214
215        let store = self.file_cache.local_store();
216        let path_provider = WriteCachePathProvider::new(self.file_cache.clone());
217        let indexer = IndexerBuilderImpl {
218            build_type: write_request.op_type.into(),
219            metadata: write_request.metadata.clone(),
220            row_group_size: write_opts.row_group_size,
221            puffin_manager: self
222                .puffin_manager_factory
223                .build(store.clone(), path_provider.clone()),
224            write_cache_enabled: true,
225            intermediate_manager: self.intermediate_manager.clone(),
226            index_options: write_request.index_options,
227            inverted_index_config: write_request.inverted_index_config,
228            fulltext_index_config: write_request.fulltext_index_config,
229            bloom_filter_index_config: write_request.bloom_filter_index_config,
230            #[cfg(feature = "vector_index")]
231            vector_index_config: write_request.vector_index_config,
232        };
233
234        let cleaner = TempFileCleaner::new(region_id, store.clone());
235        // Write to FileCache.
236        let mut writer = ParquetWriter::new_with_object_store(
237            store.clone(),
238            write_request.metadata,
239            write_request.index_config,
240            indexer,
241            path_provider.clone(),
242            metrics,
243        )
244        .await
245        .with_file_cleaner(cleaner);
246
247        let sst_info = match write_request.source {
248            either::Left(source) => {
249                writer
250                    .write_all(source, write_request.max_sequence, write_opts)
251                    .await?
252            }
253            either::Right(flat_source) => {
254                writer
255                    .write_all_flat(flat_source, write_request.max_sequence, write_opts)
256                    .await?
257            }
258        };
259
260        // Upload sst file to remote object store.
261        if sst_info.is_empty() {
262            return Ok(sst_info);
263        }
264
265        let mut upload_tracker = UploadTracker::new(region_id);
266        let mut err = None;
267        let remote_store = &upload_request.remote_store;
268        for sst in &sst_info {
269            let parquet_key = IndexKey::new(region_id, sst.file_id, FileType::Parquet);
270            let parquet_path = upload_request
271                .dest_path_provider
272                .build_sst_file_path(RegionFileId::new(region_id, sst.file_id));
273            let start = Instant::now();
274            if let Err(e) = self.upload(parquet_key, &parquet_path, remote_store).await {
275                err = Some(e);
276                break;
277            }
278            metrics.upload_parquet += start.elapsed();
279            upload_tracker.push_uploaded_file(parquet_path);
280
281            if sst.index_metadata.file_size > 0 {
282                let puffin_key = IndexKey::new(region_id, sst.file_id, FileType::Puffin(0));
283                let puffin_path = upload_request
284                    .dest_path_provider
285                    .build_index_file_path(RegionFileId::new(region_id, sst.file_id));
286                let start = Instant::now();
287                if let Err(e) = self.upload(puffin_key, &puffin_path, remote_store).await {
288                    err = Some(e);
289                    break;
290                }
291                metrics.upload_puffin += start.elapsed();
292                upload_tracker.push_uploaded_file(puffin_path);
293            }
294        }
295
296        if let Some(err) = err {
297            // Cleans files on failure.
298            upload_tracker
299                .clean(&sst_info, &self.file_cache, remote_store)
300                .await;
301            return Err(err);
302        }
303
304        Ok(sst_info)
305    }
306
307    /// Removes a file from the cache by `index_key`.
308    pub(crate) async fn remove(&self, index_key: IndexKey) {
309        self.file_cache.remove(index_key).await
310    }
311
312    /// Downloads a file in `remote_path` from the remote object store to the local cache
313    /// (specified by `index_key`).
314    pub(crate) async fn download(
315        &self,
316        index_key: IndexKey,
317        remote_path: &str,
318        remote_store: &ObjectStore,
319        file_size: u64,
320    ) -> Result<()> {
321        self.file_cache
322            .download(index_key, remote_path, remote_store, file_size)
323            .await
324    }
325
326    /// Downloads the target file into write cache only when it is not cached.
327    ///
328    /// Returns `Ok(true)` if this call performs a download, or `Ok(false)` if the
329    /// file is already present in write cache and download is skipped.
330    pub(crate) async fn download_if_absent(
331        &self,
332        index_key: IndexKey,
333        remote_path: &str,
334        remote_store: &ObjectStore,
335        file_size: u64,
336    ) -> Result<bool> {
337        if self.file_cache.contains_key(&index_key) {
338            debug!(
339                "Skip downloading file already in write cache, region: {}, file: {}",
340                index_key.region_id, index_key.file_id
341            );
342            return Ok(false);
343        }
344
345        self.download(index_key, remote_path, remote_store, file_size)
346            .await?;
347        Ok(true)
348    }
349
350    /// Uploads a Parquet file or a Puffin file to the remote object store.
351    pub(crate) async fn upload(
352        &self,
353        index_key: IndexKey,
354        upload_path: &str,
355        remote_store: &ObjectStore,
356    ) -> Result<()> {
357        let region_id = index_key.region_id;
358        let file_id = index_key.file_id;
359        let file_type = index_key.file_type;
360        let cache_path = self.file_cache.cache_file_path(index_key);
361
362        let start = Instant::now();
363        let cached_value = self
364            .file_cache
365            .local_store()
366            .stat(&cache_path)
367            .await
368            .context(error::OpenDalSnafu)?;
369        let reader = self
370            .file_cache
371            .local_store()
372            .reader(&cache_path)
373            .await
374            .context(error::OpenDalSnafu)?
375            .into_futures_async_read(0..cached_value.content_length())
376            .await
377            .context(error::OpenDalSnafu)?;
378
379        let mut writer = remote_store
380            .writer_with(upload_path)
381            .chunk(DEFAULT_WRITE_BUFFER_SIZE.as_bytes() as usize)
382            .concurrent(DEFAULT_WRITE_CONCURRENCY)
383            .await
384            .context(error::OpenDalSnafu)?
385            .into_futures_async_write();
386
387        let bytes_written =
388            futures::io::copy(reader, &mut writer)
389                .await
390                .context(error::UploadSnafu {
391                    region_id,
392                    file_id,
393                    file_type,
394                })?;
395
396        // Must close to upload all data.
397        writer.close().await.context(error::UploadSnafu {
398            region_id,
399            file_id,
400            file_type,
401        })?;
402
403        UPLOAD_BYTES_TOTAL.inc_by(bytes_written);
404
405        debug!(
406            "Successfully upload file to remote, region: {}, file: {}, upload_path: {}, cost: {:?}",
407            region_id,
408            file_id,
409            upload_path,
410            start.elapsed(),
411        );
412
413        let index_value = IndexValue {
414            file_size: bytes_written as _,
415        };
416        // Register to file cache
417        self.file_cache.put(index_key, index_value).await;
418
419        Ok(())
420    }
421
422    /// Sends a region load cache task to the background processing queue.
423    ///
424    /// If the receiver has been dropped, the error is ignored.
425    pub(crate) fn load_region_cache(&self, task: RegionLoadCacheTask) {
426        let _ = self.task_sender.send(task);
427    }
428}
429
430/// Request to write and upload a SST.
431pub struct SstUploadRequest {
432    /// Destination path provider of which SST files in write cache should be uploaded to.
433    pub dest_path_provider: RegionFilePathFactory,
434    /// Remote object store to upload.
435    pub remote_store: ObjectStore,
436}
437
438/// A structs to track files to upload and clean them if upload failed.
439pub(crate) struct UploadTracker {
440    /// Id of the region to track.
441    region_id: RegionId,
442    /// Paths of files uploaded successfully.
443    files_uploaded: Vec<String>,
444}
445
446impl UploadTracker {
447    /// Creates a new instance of `UploadTracker` for a given region.
448    pub(crate) fn new(region_id: RegionId) -> Self {
449        Self {
450            region_id,
451            files_uploaded: Vec::new(),
452        }
453    }
454
455    /// Add a file path to the list of uploaded files.
456    pub(crate) fn push_uploaded_file(&mut self, path: String) {
457        self.files_uploaded.push(path);
458    }
459
460    /// Cleans uploaded files and files in the file cache at best effort.
461    pub(crate) async fn clean(
462        &self,
463        sst_info: &SstInfoArray,
464        file_cache: &FileCacheRef,
465        remote_store: &ObjectStore,
466    ) {
467        common_telemetry::info!(
468            "Start cleaning files on upload failure, region: {}, num_ssts: {}",
469            self.region_id,
470            sst_info.len()
471        );
472
473        // Cleans files in the file cache first.
474        for sst in sst_info {
475            let parquet_key = IndexKey::new(self.region_id, sst.file_id, FileType::Parquet);
476            file_cache.remove(parquet_key).await;
477
478            if sst.index_metadata.file_size > 0 {
479                let puffin_key = IndexKey::new(
480                    self.region_id,
481                    sst.file_id,
482                    FileType::Puffin(sst.index_metadata.version),
483                );
484                file_cache.remove(puffin_key).await;
485            }
486        }
487
488        // Cleans uploaded files.
489        for file_path in &self.files_uploaded {
490            if let Err(e) = remote_store.delete(file_path).await {
491                common_telemetry::error!(e; "Failed to delete file {}", file_path);
492            }
493        }
494    }
495}
496
497#[cfg(test)]
498mod tests {
499    use bytes::Bytes;
500    use common_test_util::temp_dir::create_temp_dir;
501    use object_store::ATOMIC_WRITE_DIR;
502    use parquet::file::metadata::PageIndexPolicy;
503    use store_api::region_request::PathType;
504    use store_api::storage::FileId;
505
506    use super::*;
507    use crate::access_layer::OperationType;
508    use crate::cache::file_cache::IndexValue;
509    use crate::cache::test_util::{assert_parquet_metadata_equal, new_fs_store};
510    use crate::cache::{CacheManager, CacheStrategy};
511    use crate::error::InvalidBatchSnafu;
512    use crate::read::Source;
513    use crate::region::options::IndexOptions;
514    use crate::sst::parquet::reader::ParquetReaderBuilder;
515    use crate::test_util::TestEnv;
516    use crate::test_util::sst_util::{
517        new_batch_by_range, new_source, sst_file_handle_with_file_id, sst_region_metadata,
518    };
519
520    #[tokio::test]
521    async fn test_write_and_upload_sst() {
522        // TODO(QuenKar): maybe find a way to create some object server for testing,
523        // and now just use local file system to mock.
524        let mut env = TestEnv::new().await;
525        let mock_store = env.init_object_store_manager();
526        let path_provider = RegionFilePathFactory::new("test".to_string(), PathType::Bare);
527
528        let local_dir = create_temp_dir("");
529        let local_store = new_fs_store(local_dir.path().to_str().unwrap());
530
531        let write_cache = env
532            .create_write_cache(local_store.clone(), ReadableSize::mb(10))
533            .await;
534
535        // Create Source
536        let metadata = Arc::new(sst_region_metadata());
537        let region_id = metadata.region_id;
538        let source = new_source(&[
539            new_batch_by_range(&["a", "d"], 0, 60),
540            new_batch_by_range(&["b", "f"], 0, 40),
541            new_batch_by_range(&["b", "h"], 100, 200),
542        ]);
543
544        let write_request = SstWriteRequest {
545            op_type: OperationType::Flush,
546            metadata,
547            source: either::Left(source),
548            storage: None,
549            max_sequence: None,
550            cache_manager: Default::default(),
551            index_options: IndexOptions::default(),
552            index_config: Default::default(),
553            inverted_index_config: Default::default(),
554            fulltext_index_config: Default::default(),
555            bloom_filter_index_config: Default::default(),
556            #[cfg(feature = "vector_index")]
557            vector_index_config: Default::default(),
558        };
559
560        let upload_request = SstUploadRequest {
561            dest_path_provider: path_provider.clone(),
562            remote_store: mock_store.clone(),
563        };
564
565        let write_opts = WriteOptions {
566            row_group_size: 512,
567            ..Default::default()
568        };
569
570        // Write to cache and upload sst to mock remote store
571        let mut metrics = Metrics::new(WriteType::Flush);
572        let mut sst_infos = write_cache
573            .write_and_upload_sst(write_request, upload_request, &write_opts, &mut metrics)
574            .await
575            .unwrap();
576        let sst_info = sst_infos.remove(0);
577
578        let file_id = sst_info.file_id;
579        let sst_upload_path =
580            path_provider.build_sst_file_path(RegionFileId::new(region_id, file_id));
581        let index_upload_path =
582            path_provider.build_index_file_path(RegionFileId::new(region_id, file_id));
583
584        // Check write cache contains the key
585        let key = IndexKey::new(region_id, file_id, FileType::Parquet);
586        assert!(write_cache.file_cache.contains_key(&key));
587
588        // Check file data
589        let remote_data = mock_store.read(&sst_upload_path).await.unwrap();
590        let cache_data = local_store
591            .read(&write_cache.file_cache.cache_file_path(key))
592            .await
593            .unwrap();
594        assert_eq!(remote_data.to_vec(), cache_data.to_vec());
595
596        // Check write cache contains the index key
597        let index_key = IndexKey::new(region_id, file_id, FileType::Puffin(0));
598        assert!(write_cache.file_cache.contains_key(&index_key));
599
600        let remote_index_data = mock_store.read(&index_upload_path).await.unwrap();
601        let cache_index_data = local_store
602            .read(&write_cache.file_cache.cache_file_path(index_key))
603            .await
604            .unwrap();
605        assert_eq!(remote_index_data.to_vec(), cache_index_data.to_vec());
606
607        // Removes the file from the cache.
608        let sst_index_key = IndexKey::new(region_id, file_id, FileType::Parquet);
609        write_cache.remove(sst_index_key).await;
610        assert!(!write_cache.file_cache.contains_key(&sst_index_key));
611        write_cache.remove(index_key).await;
612        assert!(!write_cache.file_cache.contains_key(&index_key));
613    }
614
615    #[tokio::test]
616    async fn test_read_metadata_from_write_cache() {
617        common_telemetry::init_default_ut_logging();
618        let mut env = TestEnv::new().await;
619        let data_home = env.data_home().display().to_string();
620        let mock_store = env.init_object_store_manager();
621
622        let local_dir = create_temp_dir("");
623        let local_path = local_dir.path().to_str().unwrap();
624        let local_store = new_fs_store(local_path);
625
626        // Create a cache manager using only write cache
627        let write_cache = env
628            .create_write_cache(local_store.clone(), ReadableSize::mb(10))
629            .await;
630        let cache_manager = Arc::new(
631            CacheManager::builder()
632                .write_cache(Some(write_cache.clone()))
633                .build(),
634        );
635
636        // Create source
637        let metadata = Arc::new(sst_region_metadata());
638
639        let source = new_source(&[
640            new_batch_by_range(&["a", "d"], 0, 60),
641            new_batch_by_range(&["b", "f"], 0, 40),
642            new_batch_by_range(&["b", "h"], 100, 200),
643        ]);
644
645        // Write to local cache and upload sst to mock remote store
646        let write_request = SstWriteRequest {
647            op_type: OperationType::Flush,
648            metadata,
649            source: either::Left(source),
650            storage: None,
651            max_sequence: None,
652            cache_manager: cache_manager.clone(),
653            index_options: IndexOptions::default(),
654            index_config: Default::default(),
655            inverted_index_config: Default::default(),
656            fulltext_index_config: Default::default(),
657            bloom_filter_index_config: Default::default(),
658            #[cfg(feature = "vector_index")]
659            vector_index_config: Default::default(),
660        };
661        let write_opts = WriteOptions {
662            row_group_size: 512,
663            ..Default::default()
664        };
665        let upload_request = SstUploadRequest {
666            dest_path_provider: RegionFilePathFactory::new(data_home.clone(), PathType::Bare),
667            remote_store: mock_store.clone(),
668        };
669
670        let mut metrics = Metrics::new(WriteType::Flush);
671        let mut sst_infos = write_cache
672            .write_and_upload_sst(write_request, upload_request, &write_opts, &mut metrics)
673            .await
674            .unwrap();
675        let sst_info = sst_infos.remove(0);
676        let write_parquet_metadata = sst_info.file_metadata.unwrap();
677
678        // Read metadata from write cache
679        let handle = sst_file_handle_with_file_id(sst_info.file_id, 0, 1000);
680        let builder = ParquetReaderBuilder::new(
681            data_home,
682            PathType::Bare,
683            handle.clone(),
684            mock_store.clone(),
685        )
686        .cache(CacheStrategy::EnableAll(cache_manager.clone()))
687        .page_index_policy(PageIndexPolicy::Optional);
688        let reader = builder.build().await.unwrap().unwrap();
689
690        // Check parquet metadata
691        assert_parquet_metadata_equal(write_parquet_metadata, reader.parquet_metadata());
692    }
693
694    #[tokio::test]
695    async fn test_write_cache_clean_tmp_files() {
696        common_telemetry::init_default_ut_logging();
697        let mut env = TestEnv::new().await;
698        let data_home = env.data_home().display().to_string();
699        let mock_store = env.init_object_store_manager();
700
701        let write_cache_dir = create_temp_dir("");
702        let write_cache_path = write_cache_dir.path().to_str().unwrap();
703        let write_cache = env
704            .create_write_cache_from_path(write_cache_path, ReadableSize::mb(10))
705            .await;
706
707        // Create a cache manager using only write cache
708        let cache_manager = Arc::new(
709            CacheManager::builder()
710                .write_cache(Some(write_cache.clone()))
711                .build(),
712        );
713
714        // Create source
715        let metadata = Arc::new(sst_region_metadata());
716
717        // Creates a source that can return an error to abort the writer.
718        let source = Source::Iter(Box::new(
719            [
720                Ok(new_batch_by_range(&["a", "d"], 0, 60)),
721                InvalidBatchSnafu {
722                    reason: "Abort the writer",
723                }
724                .fail(),
725            ]
726            .into_iter(),
727        ));
728
729        // Write to local cache and upload sst to mock remote store
730        let write_request = SstWriteRequest {
731            op_type: OperationType::Flush,
732            metadata,
733            source: either::Left(source),
734            storage: None,
735            max_sequence: None,
736            cache_manager: cache_manager.clone(),
737            index_options: IndexOptions::default(),
738            index_config: Default::default(),
739            inverted_index_config: Default::default(),
740            fulltext_index_config: Default::default(),
741            bloom_filter_index_config: Default::default(),
742            #[cfg(feature = "vector_index")]
743            vector_index_config: Default::default(),
744        };
745        let write_opts = WriteOptions {
746            row_group_size: 512,
747            ..Default::default()
748        };
749        let upload_request = SstUploadRequest {
750            dest_path_provider: RegionFilePathFactory::new(data_home.clone(), PathType::Bare),
751            remote_store: mock_store.clone(),
752        };
753
754        let mut metrics = Metrics::new(WriteType::Flush);
755        write_cache
756            .write_and_upload_sst(write_request, upload_request, &write_opts, &mut metrics)
757            .await
758            .unwrap_err();
759        let atomic_write_dir = write_cache_dir.path().join(ATOMIC_WRITE_DIR);
760        let mut entries = tokio::fs::read_dir(&atomic_write_dir).await.unwrap();
761        let mut has_files = false;
762        while let Some(entry) = entries.next_entry().await.unwrap() {
763            if entry.file_type().await.unwrap().is_dir() {
764                continue;
765            }
766            has_files = true;
767            common_telemetry::warn!(
768                "Found remaining temporary file in atomic dir: {}",
769                entry.path().display()
770            );
771        }
772
773        assert!(!has_files);
774    }
775
776    #[tokio::test]
777    async fn test_download_if_absent_skips_when_cached() {
778        let mut env = TestEnv::new().await;
779        let remote_store = env.init_object_store_manager();
780
781        let local_dir = create_temp_dir("");
782        let local_store = new_fs_store(local_dir.path().to_str().unwrap());
783        let write_cache = env
784            .create_write_cache(local_store.clone(), ReadableSize::mb(10))
785            .await;
786
787        let region_id = RegionId::new(1024, 1);
788        let file_id = FileId::random();
789        let key = IndexKey::new(region_id, file_id, FileType::Parquet);
790        write_cache
791            .file_cache()
792            .put(key, IndexValue { file_size: 1 })
793            .await;
794
795        let downloaded = write_cache
796            .download_if_absent(key, "missing/path.parquet", &remote_store, 1)
797            .await
798            .unwrap();
799
800        assert!(!downloaded);
801    }
802
803    #[tokio::test]
804    async fn test_download_if_absent_downloads_when_missing() {
805        let mut env = TestEnv::new().await;
806        let remote_store = env.init_object_store_manager();
807
808        let local_dir = create_temp_dir("");
809        let local_store = new_fs_store(local_dir.path().to_str().unwrap());
810        let write_cache = env
811            .create_write_cache(local_store.clone(), ReadableSize::mb(10))
812            .await;
813
814        let region_id = RegionId::new(1024, 2);
815        let file_id = FileId::random();
816        let key = IndexKey::new(region_id, file_id, FileType::Parquet);
817        let remote_path = format!("download-if-absent/{file_id}.parquet");
818        let remote_data = Bytes::from_static(b"download-if-absent-test");
819        remote_store
820            .write(&remote_path, remote_data.clone())
821            .await
822            .unwrap();
823
824        let downloaded = write_cache
825            .download_if_absent(key, &remote_path, &remote_store, remote_data.len() as u64)
826            .await
827            .unwrap();
828
829        assert!(downloaded);
830        assert!(write_cache.file_cache().contains_key(&key));
831
832        let cached_data = local_store
833            .read(&write_cache.file_cache().cache_file_path(key))
834            .await
835            .unwrap();
836        assert_eq!(cached_data.to_vec(), remote_data.to_vec());
837    }
838}