[server] When applying projection pushdown, return empty records instead of skipping them to ensure offset movement in the client. #2370
+126
−50
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ead of skipping them to ensure offset movement in the client.
wuchong
reviewed
Jan 20, 2026
Member
wuchong
left a comment
wuchong
left a comment
There was a problem hiding this comment.
@loserwang1024 , I left some comments to optimize the test.
| DEFAULT_COMPRESSION); | ||
| } | ||
|
|
||
| private MemoryLogRecords logRecords( |
Comment on lines
+1172
to
+1185
| // In this way, if skip empty batch, the read will in stuck forever. | ||
| conf.set(ConfigOptions.REMOTE_LOG_TASK_INTERVAL_DURATION, Duration.ZERO); | ||
| conf.set( | ||
| ConfigOptions.LOG_SEGMENT_FILE_SIZE, | ||
| new MemorySize(5 * V0_RECORD_BATCH_HEADER_SIZE)); | ||
| conf.set( | ||
| ConfigOptions.CLIENT_SCANNER_LOG_FETCH_MAX_BYTES_FOR_BUCKET, | ||
| new MemorySize(5 * V0_RECORD_BATCH_HEADER_SIZE)); | ||
| final FlussClusterExtension flussClusterExtension = | ||
| FlussClusterExtension.builder() | ||
| .setNumOfTabletServers(3) | ||
| .setClusterConf(conf) | ||
| .build(); | ||
| flussClusterExtension.start(); |
Member
There was a problem hiding this comment.
In my local environment, the original test takes about 15 seconds, but it increases to 55 seconds after applying this PR. I think we should optimize it. Here are several optimization opportunities:
- The test starts a new Fluss cluster in addition to the existing one from the test base class, which adds significant overhead.
- The issue only occurs when projection pushdown is enabled (
doProjection = true), so there’s no need to testdoProjection = false. - The table bucket count can be reduced to 1, as it’s sufficient for reproducing the problem.
- The number of empty record batches can be reduced from 10 to 2, this should still reliably reproduce the issue while avoiding 10 unnecessary RPC round-trips that dominate the runtime.
I suggest introducing a dedicated test class like CustomFlussClusterITCase (without extending ClientToServerITCaseBase) for tests that require manual cluster management, and adding a focused test method such as testProjectionPushdownWithEmptyBatches that incorporates all these optimizations.
2 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Purpose
Linked issue: the detailed analysis is in #2369.
Brief change log
Not filter empty log in projection.
Tests
org.apache.fluss.client.table.FlussTableITCase#testFirstRowMergeEngine
API and Format
Documentation