bvaradar commented on a change in pull request #1128: [HUDI-453] Fix throw
failed to archive commits error when writing data to MOR/COW table
URL: https://github.com/apache/incubator-hudi/pull/1128#discussion_r361733653
##########
File path:
hudi-client/src/main/java/org/apache/hudi/io/HoodieCommitArchiveLog.java
##########
@@ -249,6 +249,13 @@ public void archive(List<HoodieInstant> instants) throws
HoodieCommitException {
LOG.info("Wrapper schema " + wrapperSchema.toString());
List<IndexedRecord> records = new ArrayList<>();
for (HoodieInstant hoodieInstant : instants) {
+
+ // filter empty instant, like *.commit.requested
+ byte[] instantDetails =
commitTimeline.getInstantDetails(hoodieInstant).get();
Review comment:
@lamber-ken : Instead of blindly skipping the empty metadata, can we handle
in HoodieCommitArchiveLog.convertToAvroRecord so that we still create an
archived entry but with empty commitMetadata. This way, we faithfully archive
all state transitions.
In other words, for the below two code snippets, if the state is Requested,
skip setting commit metadata but set other attributes.
case HoodieTimeline.COMMIT_ACTION: {
HoodieCommitMetadata commitMetadata = HoodieCommitMetadata
.fromBytes(commitTimeline.getInstantDetails(hoodieInstant).get(),
HoodieCommitMetadata.class);
archivedMetaWrapper.setHoodieCommitMetadata(commitMetadataConverter(commitMetadata));
archivedMetaWrapper.setActionType(ActionType.commit.name());
break;
}
case HoodieTimeline.DELTA_COMMIT_ACTION: {
HoodieCommitMetadata commitMetadata = HoodieCommitMetadata
.fromBytes(commitTimeline.getInstantDetails(hoodieInstant).get(),
HoodieCommitMetadata.class);
archivedMetaWrapper.setHoodieCommitMetadata(commitMetadataConverter(commitMetadata));
archivedMetaWrapper.setActionType(ActionType.commit.name());
break;
}
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
