[flink] Assign splits with fixed batch size in StaticFileStoreSplitEnumerator #687
+105
−21
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
JingsongLi
reviewed
Mar 22, 2023
| @@ -79,10 +82,15 @@ public void handleSplitRequest(int subtask, @Nullable String hostname) { | |||
| // The following batch assignment operation is for two purposes: | |||
| // To distribute splits evenly when batch reading to prevent a few tasks from reading all | |||
| // the data (for example, the current resource can only schedule part of the tasks). | |||
| // TODO: assignment is already created in constructor, here can just assign per batch | |||
| List<FileStoreSourceSplit> splits = pendingSplitAssignment.remove(subtask); | |||
There was a problem hiding this comment.
Can we make pendingSplitAssignment to Map<Integer, Queue<FileStoreSourceSplit>>?
And poll a batch from it.
| @@ -37,6 +37,9 @@ | |||
| public class StaticFileStoreSplitEnumerator | |||
| implements SplitEnumerator<FileStoreSourceSplit, PendingSplitsCheckpoint> { | |||
|
|
|||
| /** Default batch splits size to avoid exceed `akka.framesize`. */ | |||
| private static final int DEFAULT_SPLITS_SIZE = 10; | |||
There was a problem hiding this comment.
Can we introduce an option for this? Can be an option in FlinkConnectorOptions?
JingsongLi
reviewed
Mar 23, 2023
| @@ -26,6 +26,12 @@ | |||
| <td>String</td> | |||
| <td>The log system used to keep changes of the table.<br /><br />Possible values:<br /><ul><li>"none": No log system, the data is written only to file store, and the streaming read will be directly read from the file store.</li></ul><ul><li>"kafka": Kafka log system, the data is double written to file store and kafka, and the streaming read will be read from kafka.</li></ul></td> | |||
| </tr> | |||
| <tr> | |||
| <td><h5>scan.bounded.splits.size</h5></td> | |||
There was a problem hiding this comment.
scan.split-enumerator.batch-size?
| @@ -94,6 +98,6 @@ public SplitEnumerator<FileStoreSourceSplit, PendingSplitsCheckpoint> restoreEnu | |||
| } | |||
|
|
|||
| Snapshot snapshot = snapshotId == null ? null : snapshotManager.snapshot(snapshotId); | |||
| return new StaticFileStoreSplitEnumerator(context, snapshot, splits); | |||
| return new StaticFileStoreSplitEnumerator(context, snapshot, splits, splitsSize); | |||
There was a problem hiding this comment.
Here, we can get splitBatchSize from table. (via table.options())
| List<FileStoreSourceSplit> splits = pendingSplitAssignment.remove(subtask); | ||
| if (splits != null && splits.size() > 0) { | ||
| context.assignSplits(new SplitsAssignment<>(Collections.singletonMap(subtask, splits))); | ||
| Queue<FileStoreSourceSplit> allSubTaskSplits = pendingSplitAssignment.get(subtask); |
There was a problem hiding this comment.
minor: rename to taskSplits
| if (splits != null && splits.size() > 0) { | ||
| context.assignSplits(new SplitsAssignment<>(Collections.singletonMap(subtask, splits))); | ||
| Queue<FileStoreSourceSplit> allSubTaskSplits = pendingSplitAssignment.get(subtask); | ||
| List<FileStoreSourceSplit> batchAssignment = new ArrayList<>(); |
There was a problem hiding this comment.
minor: rename to assignment?
JingsongLi
approved these changes
Mar 24, 2023
This was referenced Mar 27, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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
Assign all splits in one time will produce exceed
akka.framesizeto exceptionsAdd
scan.bounded.splits.sizeconfig option to detemine how many splits will be assign per batchTests
StaticFileStoreSplitEnumeratorTest.testSplitBatch