sankarh commented on a change in pull request #579: HIVE-21109 : Support stats replication for ACID tables. URL: https://github.com/apache/hive/pull/579#discussion_r269098036
########## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/DDLTask.java ########## @@ -2689,7 +2689,19 @@ private int alterTable(Hive db, AlterTableDesc alterTbl) throws HiveException { } else { // Note: this is necessary for UPDATE_STATISTICS command, that operates via ADDPROPS (why?). // For any other updates, we don't want to do txn check on partitions when altering table. - boolean isTxn = alterTbl.getPartSpec() != null && alterTbl.getOp() == AlterTableTypes.ADDPROPS; + boolean isTxn = false; + if (alterTbl.getPartSpec() != null && alterTbl.getOp() == AlterTableTypes.ADDPROPS) { + // ADDPROPS is used to add repl.last.id during replication. That's not a transactional + // change. + Map<String, String> props = alterTbl.getProps(); + if (props.size() <= 1 && props.get(ReplicationSpec.KEY.CURR_STATE_ID.toString()) != null) { + isTxn = false; + } else { + isTxn = true; + } + } + // TODO: Somehow we have to signal alterPartitions that it's part of replication and + // should use replication's valid writeid list instead of creating one. Review comment: What do you mean by replication's valid writeid list in this comment? Even in repl flow, we get validWriteIdList from HMS based on incoming writeId in the event msg. Are you suggesting to cache this ValidWriteIdList somewhere and use it instead of invoking HMS API? ---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services