aasha commented on a change in pull request #951: HIVE-22997 : Copy external
table to target during Repl Dump operation
URL: https://github.com/apache/hive/pull/951#discussion_r394297682
##########
File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/ReplDumpTask.java
##########
@@ -662,17 +690,48 @@ void dumpTable(String dbName, String tblName, String
validTxnList, Path dbRoot,
replLogger.tableLog(tblName, tableSpec.tableHandle.getTableType());
if (tableSpec.tableHandle.getTableType().equals(TableType.EXTERNAL_TABLE)
|| Utils.shouldDumpMetaDataOnly(conf)) {
- return;
+ return Collections.EMPTY_LIST;
+ }
+ return replPathMappings;
+ }
+
+ private void intitiateDataCopyTasks() {
+ Iterator<ExternalTableCopyTaskBuilder.DirCopyWork> extCopyWorkItr =
work.getDirCopyIterator();
+ ReplOperationCompleteAckWork replDumpCompleteAckWork = new
ReplOperationCompleteAckWork(dumpAckFile);
+ Task<ReplOperationCompleteAckWork> dumpCompleteAckWorkTask =
TaskFactory.get(replDumpCompleteAckWork, conf);
+ List<Task<?>> childTasks = new ArrayList<>();
+ int maxTasks =
conf.getIntVar(HiveConf.ConfVars.REPL_APPROX_MAX_LOAD_TASKS);
+ TaskTracker taskTracker = new TaskTracker(maxTasks);
+ while (taskTracker.canAddMoreTasks() && hasMoreCopyWork()) {
+ if (work.replPathIteratorInitialized() && extCopyWorkItr.hasNext()) {
+ childTasks.addAll(new ExternalTableCopyTaskBuilder(work,
conf).tasks(taskTracker));
+ } else {
+ childTasks.addAll(ReplPathMapping.tasks(work, taskTracker, conf));
+ }
}
- for (ReplPathMapping replPathMapping: replPathMappings) {
- Task<?> copyTask = ReplCopyTask.getLoadCopyTask(
- tuple.replicationSpec, replPathMapping.getSrcPath(),
replPathMapping.getTargetPath(), conf, false);
- this.addDependentTask(copyTask);
- LOG.info("Scheduled a repl copy task from [{}] to [{}]",
- replPathMapping.getSrcPath(), replPathMapping.getTargetPath());
+ if (!childTasks.isEmpty()) {
+ boolean ackTaskAdded = false;
+ if (taskTracker.canAddMoreTasks()) {
+ childTasks.add(dumpCompleteAckWorkTask);
+ ackTaskAdded = true;
+ }
+ if (hasMoreCopyWork() || !ackTaskAdded) {
+ DAGTraversal.traverse(childTasks, new
AddDependencyToLeaves(TaskFactory.get(work, conf)));
Review comment:
Add the ack only after no more work is pending
----------------------------------------------------------------
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]
