danny0405 commented on code in PR #4336:
URL: https://github.com/apache/hudi/pull/4336#discussion_r847271083
##########
hudi-client/hudi-client-common/src/main/java/org/apache/hudi/metadata/HoodieBackedTableMetadataWriter.java:
##########
@@ -706,7 +706,20 @@ protected void
compactIfNecessary(AbstractHoodieWriteClient writeClient, String
}
}
- protected void doClean(AbstractHoodieWriteClient writeClient, String
instantTime) {
+ protected void cleanIfNecessary(AbstractHoodieWriteClient writeClient,
String instantTime) {
+ Option<HoodieInstant> lastCompletedCompactionInstant =
metadataMetaClient.reloadActiveTimeline()
+ .getCommitTimeline().filterCompletedInstants().lastInstant();
+ if (lastCompletedCompactionInstant.isPresent()
+ && metadataMetaClient.getActiveTimeline().filterCompletedInstants()
+
.findInstantsAfter(lastCompletedCompactionInstant.get().getTimestamp()).countInstants()
< 3) {
+ // do not clean the log files immediately after compaction to give some
buffer time for metadata table reader,
+ // because there is case that the reader has prepared for the log file
readers already before the compaction completes
+ // while before/during the reading of the log files, the cleaning
triggers and delete the reading files,
+ // then a FileNotFoundException(for LogFormatReader) or NPE(for
HFileReader) would throw.
+
+ // 3 is a value that I think is enough for metadata table reader.
Review Comment:
Just a subjective value that i think the metadata table reads fast enough
within 3 transaction interval. Do you have better suggestion ?
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
