Skip to content

[core] Optimize overwrite commit to use CommitResult to retry #4661

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
tsreaper merged 1 commit into from
Dec 9, 2024
Merged

[core] Optimize overwrite commit to use CommitResult to retry #4661

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
109 changes: 42 additions & 67 deletions paimon-core/src/main/java/org/apache/paimon/operation/FileStoreCommitImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -747,9 +747,7 @@ private int tryCommit(
retryResult = (RetryResult) result;

if (retryCount >= commitMaxRetries) {
if (retryResult != null) {
retryResult.cleanAll();
}
retryResult.cleanAll();
throw new RuntimeException(
String.format(
"Commit failed after %s retries, there maybe exist commit conflicts between multiple jobs.",
Expand All @@ -767,75 +765,52 @@ private int tryOverwrite(
long identifier,
@Nullable Long watermark,
Map<Integer, Long> logOffsets) {
int retryCount = 0;
while (true) {
Snapshot latestSnapshot = snapshotManager.latestSnapshot();

List<ManifestEntry> changesWithOverwrite = new ArrayList<>();
List<IndexManifestEntry> indexChangesWithOverwrite = new ArrayList<>();
if (latestSnapshot != null) {
List<ManifestEntry> currentEntries =
scan.withSnapshot(latestSnapshot)
.withPartitionFilter(partitionFilter)
.withKind(ScanMode.ALL)
.plan()
.files();
for (ManifestEntry entry : currentEntries) {
changesWithOverwrite.add(
new ManifestEntry(
FileKind.DELETE,
entry.partition(),
entry.bucket(),
entry.totalBuckets(),
entry.file()));
}
// collect all files with overwrite
Snapshot latestSnapshot = snapshotManager.latestSnapshot();
List<ManifestEntry> changesWithOverwrite = new ArrayList<>();
List<IndexManifestEntry> indexChangesWithOverwrite = new ArrayList<>();
if (latestSnapshot != null) {
List<ManifestEntry> currentEntries =
scan.withSnapshot(latestSnapshot)
.withPartitionFilter(partitionFilter)
.withKind(ScanMode.ALL)
.plan()
.files();
for (ManifestEntry entry : currentEntries) {
changesWithOverwrite.add(
new ManifestEntry(
FileKind.DELETE,
entry.partition(),
entry.bucket(),
entry.totalBuckets(),
entry.file()));
}

// collect index files
if (latestSnapshot.indexManifest() != null) {
List<IndexManifestEntry> entries =
indexManifestFile.read(latestSnapshot.indexManifest());
for (IndexManifestEntry entry : entries) {
if (partitionFilter == null || partitionFilter.test(entry.partition())) {
indexChangesWithOverwrite.add(entry.toDeleteEntry());
}
// collect index files
if (latestSnapshot.indexManifest() != null) {
List<IndexManifestEntry> entries =
indexManifestFile.read(latestSnapshot.indexManifest());
for (IndexManifestEntry entry : entries) {
if (partitionFilter == null || partitionFilter.test(entry.partition())) {
indexChangesWithOverwrite.add(entry.toDeleteEntry());
}
}
}
changesWithOverwrite.addAll(changes);
indexChangesWithOverwrite.addAll(indexFiles);

CommitResult result =
tryCommitOnce(
null,
changesWithOverwrite,
Collections.emptyList(),
indexChangesWithOverwrite,
identifier,
watermark,
logOffsets,
Snapshot.CommitKind.OVERWRITE,
latestSnapshot,
mustConflictCheck(),
branchName,
null);

if (result.isSuccess()) {
break;
}

// TODO optimize OVERWRITE too
RetryResult retryResult = (RetryResult) result;
retryResult.cleanAll();

if (retryCount >= commitMaxRetries) {
throw new RuntimeException(
String.format(
"Commit failed after %s retries, there maybe exist commit conflicts between multiple jobs.",
commitMaxRetries));
}
retryCount++;
}
return retryCount + 1;
changesWithOverwrite.addAll(changes);
indexChangesWithOverwrite.addAll(indexFiles);

return tryCommit(
changesWithOverwrite,
Collections.emptyList(),
indexChangesWithOverwrite,
identifier,
watermark,
logOffsets,
Snapshot.CommitKind.OVERWRITE,
mustConflictCheck(),
branchName,
null);
}

@VisibleForTesting
Expand Down
Loading