[ 
https://issues.apache.org/jira/browse/HIVE-26894?focusedWorklogId=837200&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-837200
 ]

ASF GitHub Bot logged work on HIVE-26894:
-----------------------------------------

                Author: ASF GitHub Bot
            Created on: 05/Jan/23 10:30
            Start Date: 05/Jan/23 10:30
    Worklog Time Spent: 10m 
      Work Description: sankarh commented on code in PR #3909:
URL: https://github.com/apache/hive/pull/3909#discussion_r1062211047


##########
ql/src/java/org/apache/hadoop/hive/ql/metadata/Hive.java:
##########
@@ -5263,23 +5263,38 @@ public Void call() throws HiveException {
 
   private static void moveAcidFilesForDelta(String deltaFileType, FileSystem 
fs,
                                             Path dst, Set<Path> 
createdDeltaDirs,
-                                            List<Path> newFiles, FileStatus 
deltaStat) throws HiveException {
-
+                                            List<Path> newFiles, FileStatus 
deltaStat, HiveConf conf) throws HiveException {
+    String configuredOwner = HiveConf.getVar(conf, 
ConfVars.HIVE_LOAD_DATA_OWNER);
     Path deltaPath = deltaStat.getPath();
     // Create the delta directory.  Don't worry if it already exists,
     // as that likely means another task got to it first.  Then move each of 
the buckets.
     // it would be more efficient to try to move the delta with it's buckets 
but that is
     // harder to make race condition proof.
     Path deltaDest = new Path(dst, deltaPath.getName());
     try {
+      FileSystem destFs = deltaDest.getFileSystem(conf);
       if (!createdDeltaDirs.contains(deltaDest)) {
         try {
-          if(fs.mkdirs(deltaDest)) {
+          // Check if the src and dest filesystems are same or not
+          // if the src and dest filesystems are different, then we need to do 
copy, instead of rename
+          if (needToCopy(conf, deltaStat.getPath(), deltaDest, fs, destFs, 
configuredOwner, true)) {

Review Comment:
   Can you add tests that cover these 2 newly added flows?



##########
ql/src/java/org/apache/hadoop/hive/ql/metadata/Hive.java:
##########
@@ -5263,23 +5263,38 @@ public Void call() throws HiveException {
 
   private static void moveAcidFilesForDelta(String deltaFileType, FileSystem 
fs,
                                             Path dst, Set<Path> 
createdDeltaDirs,
-                                            List<Path> newFiles, FileStatus 
deltaStat) throws HiveException {
-
+                                            List<Path> newFiles, FileStatus 
deltaStat, HiveConf conf) throws HiveException {
+    String configuredOwner = HiveConf.getVar(conf, 
ConfVars.HIVE_LOAD_DATA_OWNER);
     Path deltaPath = deltaStat.getPath();
     // Create the delta directory.  Don't worry if it already exists,
     // as that likely means another task got to it first.  Then move each of 
the buckets.
     // it would be more efficient to try to move the delta with it's buckets 
but that is
     // harder to make race condition proof.
     Path deltaDest = new Path(dst, deltaPath.getName());
     try {
+      FileSystem destFs = deltaDest.getFileSystem(conf);
       if (!createdDeltaDirs.contains(deltaDest)) {
         try {
-          if(fs.mkdirs(deltaDest)) {
+          // Check if the src and dest filesystems are same or not
+          // if the src and dest filesystems are different, then we need to do 
copy, instead of rename
+          if (needToCopy(conf, deltaStat.getPath(), deltaDest, fs, destFs, 
configuredOwner, true)) {
+            //copy if across file system or encryption zones.
+            LOG.debug("Copying source " + deltaStat.getPath() + " to " + 
deltaDest + " because HDFS encryption zones are different.");

Review Comment:
   The log msg can also highlight, the copy is across different file systems or 
encryption zones.



##########
ql/src/java/org/apache/hadoop/hive/ql/metadata/Hive.java:
##########
@@ -5294,11 +5309,24 @@ private static void moveAcidFilesForDelta(String 
deltaFileType, FileSystem fs,
       for (FileStatus bucketStat : bucketStats) {
         Path bucketSrc = bucketStat.getPath();
         Path bucketDest = new Path(deltaDest, bucketSrc.getName());
+        FileSystem bucketDestFs = bucketDest.getFileSystem(conf);
         LOG.info("Moving bucket " + bucketSrc.toUri().toString() + " to " +
                 bucketDest.toUri().toString());
         try {
-          fs.rename(bucketSrc, bucketDest);
-          if (newFiles != null) {
+            // Check if the src and dest filesystems are same or not
+            // if the src and dest filesystems are different, then we need to 
do copy, instead of rename
+            if (needToCopy(conf, bucketSrc, bucketDest, fs, bucketDestFs, 
configuredOwner, true)) {
+              //copy if across file system or encryption zones.
+              LOG.debug("Copying source " + bucketSrc + " to " + bucketDest + 
" because HDFS encryption zones are different.");
+              FileUtils.copy(fs, bucketSrc, bucketDestFs, bucketDest,
+                   true,    // delete source
+                   true, // overwrite destination
+                   conf);
+            } else {
+                // do the rename
+                bucketDestFs.rename(bucketSrc, bucketDest);
+            }
+            if (newFiles != null) {
             newFiles.add(bucketDest);

Review Comment:
   Need to add extra tab space for the body of "if" path.





Issue Time Tracking
-------------------

    Worklog Id:     (was: 837200)
    Time Spent: 1h  (was: 50m)

> After using scratchdir for staging final job, CTAS and IOW on ACID tables are 
> failing with wrongFS exception
> ------------------------------------------------------------------------------------------------------------
>
>                 Key: HIVE-26894
>                 URL: https://issues.apache.org/jira/browse/HIVE-26894
>             Project: Hive
>          Issue Type: Sub-task
>          Components: Hive
>    Affects Versions: 3.1.3
>            Reporter: Sruthi M
>            Assignee: Sruthi M
>            Priority: Major
>              Labels: pull-request-available
>          Time Spent: 1h
>  Remaining Estimate: 0h
>
> ERROR : Failed with exception Wrong FS: 
> abfs://xxxxxxxx/hive/warehouse/managed/tpcds_orc.db/test_sales/delta_0000001_0000001_0000,
>  expected: hdfs://mycluster



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to