pkumarsinha 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_r395603150
##########
File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/ReplDumpTask.java
##########
@@ -662,15 +696,37 @@ 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;
}
- 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());
+ return replPathMappings;
+ }
+
+ private void intitiateDataCopyTasks() throws SemanticException {
+ Iterator<ExternalTableCopyTaskBuilder.DirCopyWork> extCopyWorkItr =
work.getDirCopyIterator();
+ 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));
+ }
+ }
+ if (!childTasks.isEmpty()) {
+ DAGTraversal.traverse(childTasks, new
AddDependencyToLeaves(TaskFactory.get(work, conf)));
+ } else {
+ prepareReturnValues(work.getResultValues());
+ childTasks.add(TaskFactory.get(new
ReplOperationCompleteAckWork(work.getDumpAckFile()), conf));
}
Review comment:
Yes, in my previous patch, I was adding the ack work as child task whenever
it was possible by making use of tracker.canAddMoreTasks . Then the requirement
came to run this prepareReturnValues(); as the last operation just before ack
task.
----------------------------------------------------------------
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]
