ajfabbri commented on a change in pull request #951: HADOOP-15183. S3Guard store becomes inconsistent after partial failure of rename URL: https://github.com/apache/hadoop/pull/951#discussion_r293155215
########## File path: hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/s3guard/MetadataStore.java ########## @@ -169,15 +186,33 @@ void move(Collection<Path> pathsToDelete, @RetryTranslated void put(PathMetadata meta) throws IOException; + /** + * Saves metadata for exactly one path, potentially + * using any bulk operation state to eliminate duplicate work. + * Review comment: So, if `operationState` is not null, can implementations may defer write to metastore until a later time, or must they write immediately but are allowed to elide subsequent "duplicate" writes? I don't think the "when is it durable" contract is super important currently, but might be worth clarifying if you roll another version of the patch. ---------------------------------------------------------------- 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 --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
