[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=473732=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-473732 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 24/Aug/20 04:24 Start Date: 24/Aug/20 04:24 Worklog Time Spent: 10m Work Description: pkumarsinha closed pull request #1084: URL: https://github.com/apache/hive/pull/1084 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 Issue Time Tracking --- Worklog Id: (was: 473732) Time Spent: 2h 50m (was: 2h 40m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 2h 50m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=473715=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-473715 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 24/Aug/20 00:41 Start Date: 24/Aug/20 00:41 Worklog Time Spent: 10m Work Description: github-actions[bot] commented on pull request #1084: URL: https://github.com/apache/hive/pull/1084#issuecomment-678849350 This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Feel free to reach out on the d...@hive.apache.org list if the patch is in need of reviews. 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 Issue Time Tracking --- Worklog Id: (was: 473715) Time Spent: 2h 40m (was: 2.5h) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 2h 40m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=446576=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-446576 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 16/Jun/20 15:13 Start Date: 16/Jun/20 15:13 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r440929491 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); +Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Task movePartitionTask = null; if (loadFileType != LoadFileType.IGNORE) { // no need to create move task, if file is moved directly to target location. - movePartitionTask = movePartitionTask(table, partSpec, stagingDir, loadFileType); + movePartitionTask = movePartitionTask(table, partSpec, moveSource, loadFileType); } - -if (ptnRootTask == null) { - ptnRootTask = copyTask; +if (performOnlyMove) { + if (ptnRootTask == null) { +ptnRootTask = addPartTask; Review comment: Not changing anything with respect to addPartTask. Just the copy-move is getting replaced with single move. 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 Issue Time Tracking --- Worklog Id: (was: 446576) Time Spent: 2.5h (was: 2h 20m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 2.5h > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=446571=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-446571 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 16/Jun/20 15:03 Start Date: 16/Jun/20 15:03 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r440922085 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -320,7 +334,7 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc private String getPartitionName(Path partitionMetadataFullPath) { //Get partition name by removing the metadata base path. //Needed for getting the data path -return partitionMetadataFullPath.toString().substring(event.metadataPath().toString().length()); +return partitionMetadataFullPath.toString().substring(event.metadataPath().toString().length() + 1); Review comment: Existing method. just changing the index 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 Issue Time Tracking --- Worklog Id: (was: 446571) Time Spent: 2h 20m (was: 2h 10m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 2h 20m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=446569=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-446569 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 16/Jun/20 15:02 Start Date: 16/Jun/20 15:02 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r440921354 ## File path: itests/hive-unit/src/test/java/org/apache/hadoop/hive/ql/parse/TestReplicationScenariosAcidTables.java ## @@ -1596,6 +1595,10 @@ public void testCheckPointingWithNewTablesAdded() throws Throwable { .run("insert into t3 values (3)") .dump(primaryDbName, dumpClause); +assertEquals(modifiedTimeTable1, fs.getFileStatus(tablet1Path).getModificationTime()); Review comment: Same as above 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 Issue Time Tracking --- Worklog Id: (was: 446569) Time Spent: 2h 10m (was: 2h) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 2h 10m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=446568=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-446568 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 16/Jun/20 15:02 Start Date: 16/Jun/20 15:02 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r440921202 ## File path: itests/hive-unit/src/test/java/org/apache/hadoop/hive/ql/parse/TestReplicationScenariosAcidTables.java ## @@ -1539,15 +1539,14 @@ public void testCheckPointingWithSourceTableDataInserted() throws Throwable { .run("insert into t2 values (24)") .run("insert into t1 values (4)") .dump(primaryDbName, dumpClause); - +assertEquals(modifiedTimeTable1CopyFile, fs.listStatus(tablet1Path)[0].getModificationTime()); Review comment: because data is moved now. We are checking this late 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 Issue Time Tracking --- Worklog Id: (was: 446568) Time Spent: 2h (was: 1h 50m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 2h > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=446567=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-446567 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 16/Jun/20 15:01 Start Date: 16/Jun/20 15:01 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r440920857 ## File path: itests/hive-unit/src/test/java/org/apache/hadoop/hive/ql/parse/TestReplicationScenariosExclusiveReplica.java ## @@ -179,6 +179,63 @@ public void externalTableReplicationWithLocalStaging() throws Throwable { .verifyResult("800"); } + @Test Review comment: Comment not clear 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 Issue Time Tracking --- Worklog Id: (was: 446567) Time Spent: 1h 50m (was: 1h 40m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 1h 50m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=446527=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-446527 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 16/Jun/20 13:53 Start Date: 16/Jun/20 13:53 Worklog Time Spent: 10m Work Description: aasha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r440855768 ## File path: itests/hive-unit/src/test/java/org/apache/hadoop/hive/ql/parse/TestReplicationScenariosAcidTables.java ## @@ -1539,15 +1539,14 @@ public void testCheckPointingWithSourceTableDataInserted() throws Throwable { .run("insert into t2 values (24)") .run("insert into t1 values (4)") .dump(primaryDbName, dumpClause); - +assertEquals(modifiedTimeTable1CopyFile, fs.listStatus(tablet1Path)[0].getModificationTime()); Review comment: Why is this changed? ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); +Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Task movePartitionTask = null; if (loadFileType != LoadFileType.IGNORE) { // no need to create move task, if file is moved directly to target location. - movePartitionTask = movePartitionTask(table, partSpec, stagingDir, loadFileType); + movePartitionTask = movePartitionTask(table, partSpec, moveSource, loadFileType); } - -if (ptnRootTask == null) { - ptnRootTask = copyTask; +if (performOnlyMove) { + if (ptnRootTask == null) { +ptnRootTask = addPartTask; Review comment: How was addPartTask added before your change? ## File path: itests/hive-unit/src/test/java/org/apache/hadoop/hive/ql/parse/TestReplicationScenariosExclusiveReplica.java ## @@ -179,6 +179,63 @@ public void externalTableReplicationWithLocalStaging() throws Throwable { .verifyResult("800"); } + @Test + public void testHdfsMoveOptimizationOnTargetStaging() throws Throwable { Review comment: Check for empty staging as the data is moved. ## File path: itests/hive-unit/src/test/java/org/apache/hadoop/hive/ql/parse/TestReplicationScenariosAcidTables.java ## @@ -1596,6 +1595,10 @@ public void testCheckPointingWithNewTablesAdded() throws Throwable { .run("insert into t3 values (3)") .dump(primaryDbName, dumpClause); +assertEquals(modifiedTimeTable1, fs.getFileStatus(tablet1Path).getModificationTime()); Review comment: Why is this changed? ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -320,7 +334,7 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc private String getPartitionName(Path partitionMetadataFullPath) { //Get partition name by removing the metadata base path. //Needed for getting the data path -return partitionMetadataFullPath.toString().substring(event.metadataPath().toString().length()); +return partitionMetadataFullPath.toString().substring(event.metadataPath().toString().length() + 1); Review comment: Can we do a split instead at '/'. This may be error prone ## File path: ql/src/java/org/apache/hadoop/hive/ql/parse/ImportSemanticAnalyzer.java ## @@ -686,9 +700,25 @@ private static ImportTableDesc getBaseCreateTableDescFromTable(String dbName, loadTableWork.setInheritTableSpecs(false); moveWork.setLoadTableWork(loadTableWork); } + Task loadPartTask = TaskFactory.get(moveWork, x.getConf()); + if (performOnlyMove) { +if (addPartTask != null) { + addPartTask.addDependentTask(loadPartTask); +} +x.getTasks().add(loadPartTask); +return addPartTask == null ? loadPartTask : addPartTask; + } + + Task copyTask = null; + if (replicationSpec.isInReplicationScope()) { Review comment: This check is already done before. Can be simplified ## File path:
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444307=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444307 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:55 Start Date: 11/Jun/20 14:55 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438847033 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); +Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Task movePartitionTask = null; if (loadFileType != LoadFileType.IGNORE) { // no need to create move task, if file is moved directly to target location. - movePartitionTask = movePartitionTask(table, partSpec, stagingDir, loadFileType); + movePartitionTask = movePartitionTask(table, partSpec, moveSource, loadFileType); } - -if (ptnRootTask == null) { - ptnRootTask = copyTask; +if (performOnlyMove) { + if (ptnRootTask == null) { +ptnRootTask = addPartTask; Review comment: No, that happens only when isOnlyDDLOperation is true. It doesn't come down in to this line in that case. 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 Issue Time Tracking --- Worklog Id: (was: 444307) Time Spent: 1.5h (was: 1h 20m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 1.5h > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444304=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444304 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:50 Start Date: 11/Jun/20 14:50 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438843549 ## File path: ql/src/java/org/apache/hadoop/hive/ql/parse/ImportSemanticAnalyzer.java ## @@ -452,7 +453,7 @@ private static ImportTableDesc getBaseCreateTableDescFromTable(String dbName, boolean isAutoPurge = false; boolean needRecycle = false; boolean copyToMigratedTxnTable = replicationSpec.isMigratingToTxnTable(); - +boolean performOnlyMove = replicationSpec.isInReplicationScope() && Utils.onSameHDFSFileSystem(dataPath, tgtPath); Review comment: As we would like to use it only in replication cases. 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 Issue Time Tracking --- Worklog Id: (was: 444304) Time Spent: 1h 20m (was: 1h 10m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 1h 20m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444303=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444303 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:48 Start Date: 11/Jun/20 14:48 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438842356 ## File path: ql/src/java/org/apache/hadoop/hive/ql/parse/ImportSemanticAnalyzer.java ## @@ -686,9 +687,25 @@ private static ImportTableDesc getBaseCreateTableDescFromTable(String dbName, loadTableWork.setInheritTableSpecs(false); moveWork.setLoadTableWork(loadTableWork); } + Task loadPartTask = TaskFactory.get(moveWork, x.getConf()); + if (performOnlyMove) { +if (addPartTask != null) { + addPartTask.addDependentTask(loadPartTask); +} +x.getTasks().add(loadPartTask); +return addPartTask == null ? loadPartTask : addPartTask; Review comment: Yes, but then we will have to have two return statements which I wanted to avoid. 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 Issue Time Tracking --- Worklog Id: (was: 444303) Time Spent: 1h 10m (was: 1h) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 1h 10m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444301=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444301 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:46 Start Date: 11/Jun/20 14:46 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438840991 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadTable.java ## @@ -297,17 +298,15 @@ static TableLocationTuple tableLocation(ImportTableDesc tblDesc, Database parent LOG.debug("adding dependent CopyWork/AddPart/MoveWork for table " + table.getCompleteName() + " with source location: " + dataPath.toString() + " and target location " + tgtPath.toString()); - -Task copyTask = ReplCopyTask.getLoadCopyTask(replicationSpec, dataPath, tmpPath, context.hiveConf, -false, false); +Path moveSrcPath = performOnlyMove ? dataPath : tmpPath; Review comment: moveSrcPath is the path used to perform move data to final warehouse dir on target. 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 Issue Time Tracking --- Worklog Id: (was: 444301) Time Spent: 1h (was: 50m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 1h > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444297=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444297 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:44 Start Date: 11/Jun/20 14:44 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438839329 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); +Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Task movePartitionTask = null; if (loadFileType != LoadFileType.IGNORE) { // no need to create move task, if file is moved directly to target location. - movePartitionTask = movePartitionTask(table, partSpec, stagingDir, loadFileType); + movePartitionTask = movePartitionTask(table, partSpec, moveSource, loadFileType); Review comment: Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Move src is either partDataSrc(which is repl Staging dir) or load's own local staging dir which is used to do operation, 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 Issue Time Tracking --- Worklog Id: (was: 444297) Time Spent: 40m (was: 0.5h) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 40m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444298=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444298 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:44 Start Date: 11/Jun/20 14:44 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438839329 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); +Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Task movePartitionTask = null; if (loadFileType != LoadFileType.IGNORE) { // no need to create move task, if file is moved directly to target location. - movePartitionTask = movePartitionTask(table, partSpec, stagingDir, loadFileType); + movePartitionTask = movePartitionTask(table, partSpec, moveSource, loadFileType); Review comment: Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Move src is either partDataSrc(which is repl Staging dir) or load's own local staging dir which is used to do data copy operation. 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 Issue Time Tracking --- Worklog Id: (was: 444298) Time Spent: 50m (was: 40m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 50m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444295=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444295 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 14:40 Start Date: 11/Jun/20 14:40 Worklog Time Spent: 10m Work Description: pkumarsinha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438836524 ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); Review comment: Normally yes, but in this case it will not as getPartitionName(sourceWarehousePartitionLocation) returns value with / attached. Path will take that itself as final path. 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 Issue Time Tracking --- Worklog Id: (was: 444295) Time Spent: 0.5h (was: 20m) > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Labels: pull-request-available > Attachments: HIVE-23539.01.patch > > Time Spent: 0.5h > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=444265=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-444265 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 11/Jun/20 13:36 Start Date: 11/Jun/20 13:36 Worklog Time Spent: 10m Work Description: aasha commented on a change in pull request #1084: URL: https://github.com/apache/hive/pull/1084#discussion_r438162512 ## File path: ql/src/java/org/apache/hadoop/hive/ql/parse/repl/dump/Utils.java ## @@ -367,6 +369,17 @@ public static boolean shouldReplicate(NotificationEvent tableForEvent, } } + public static boolean onSameHDFSFileSystem(Path dataPath, Path tgtPath) { Review comment: Unit tests for this method ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); Review comment: Can create a path directly from the constructor of Path instead of using File Separator ## File path: ql/src/java/org/apache/hadoop/hive/ql/parse/ImportSemanticAnalyzer.java ## @@ -686,9 +687,25 @@ private static ImportTableDesc getBaseCreateTableDescFromTable(String dbName, loadTableWork.setInheritTableSpecs(false); moveWork.setLoadTableWork(loadTableWork); } + Task loadPartTask = TaskFactory.get(moveWork, x.getConf()); + if (performOnlyMove) { +if (addPartTask != null) { + addPartTask.addDependentTask(loadPartTask); +} +x.getTasks().add(loadPartTask); +return addPartTask == null ? loadPartTask : addPartTask; Review comment: can be simplified. There is already a not null check above ## File path: ql/src/java/org/apache/hadoop/hive/ql/parse/ImportSemanticAnalyzer.java ## @@ -452,7 +453,7 @@ private static ImportTableDesc getBaseCreateTableDescFromTable(String dbName, boolean isAutoPurge = false; boolean needRecycle = false; boolean copyToMigratedTxnTable = replicationSpec.isMigratingToTxnTable(); - +boolean performOnlyMove = replicationSpec.isInReplicationScope() && Utils.onSameHDFSFileSystem(dataPath, tgtPath); Review comment: why do we need this extra condition replicationSpec.isInReplicationScope()? ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -243,29 +246,34 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc : LoadFileType.OVERWRITE_EXISTING); stagingDir = PathUtils.getExternalTmpPath(replicaWarehousePartitionLocation, context.pathInfo); } - -Task copyTask = ReplCopyTask.getLoadCopyTask( -event.replicationSpec(), -new Path(event.dataPath() + Path.SEPARATOR + getPartitionName(sourceWarehousePartitionLocation)), -stagingDir, -context.hiveConf, false, false -); - +Path partDataSrc = new Path(event.dataPath() + File.separator + getPartitionName(sourceWarehousePartitionLocation)); +Path moveSource = performOnlyMove ? partDataSrc : stagingDir; Task movePartitionTask = null; if (loadFileType != LoadFileType.IGNORE) { // no need to create move task, if file is moved directly to target location. - movePartitionTask = movePartitionTask(table, partSpec, stagingDir, loadFileType); + movePartitionTask = movePartitionTask(table, partSpec, moveSource, loadFileType); } - -if (ptnRootTask == null) { - ptnRootTask = copyTask; +if (performOnlyMove) { + if (ptnRootTask == null) { +ptnRootTask = addPartTask; Review comment: will the addPartTask not be added already as part of DDL operation? ## File path: ql/src/java/org/apache/hadoop/hive/ql/exec/repl/bootstrap/load/table/LoadPartitions.java ## @@ -225,6 +227,7 @@ private void addPartition(boolean hasMorePartitions, AlterTableAddPartitionDesc } Path stagingDir = replicaWarehousePartitionLocation; +boolean performOnlyMove = Utils.onSameHDFSFileSystem(event.dataPath(), replicaWarehousePartitionLocation); Review comment: can use stagingDir instead of
[jira] [Work logged] (HIVE-23539) Optimize data copy during repl load operation for HDFS based staging location
[ https://issues.apache.org/jira/browse/HIVE-23539?focusedWorklogId=443269=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-443269 ] ASF GitHub Bot logged work on HIVE-23539: - Author: ASF GitHub Bot Created on: 09/Jun/20 16:48 Start Date: 09/Jun/20 16:48 Worklog Time Spent: 10m Work Description: pkumarsinha opened a new pull request #1084: URL: https://github.com/apache/hive/pull/1084 …ased staging location ## NOTICE Please create an issue in ASF JIRA before opening a pull request, and you need to set the title of the pull request which starts with the corresponding JIRA issue number. (e.g. HIVE-X: Fix a typo in YYY) For more details, please see https://cwiki.apache.org/confluence/display/Hive/HowToContribute 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 Issue Time Tracking --- Worklog Id: (was: 443269) Remaining Estimate: 0h Time Spent: 10m > Optimize data copy during repl load operation for HDFS based staging location > - > > Key: HIVE-23539 > URL: https://issues.apache.org/jira/browse/HIVE-23539 > Project: Hive > Issue Type: Improvement >Reporter: Pravin Sinha >Assignee: Pravin Sinha >Priority: Major > Attachments: HIVE-23539.01.patch > > Time Spent: 10m > Remaining Estimate: 0h > -- This message was sent by Atlassian Jira (v8.3.4#803005)