kevinjmh commented on a change in pull request #3780:
URL: https://github.com/apache/carbondata/pull/3780#discussion_r432950129



##########
File path: integration/spark/src/main/scala/org/apache/spark/sql/CarbonEnv.scala
##########
@@ -328,7 +328,7 @@ object CarbonEnv {
     if ((!EnvHelper.isLegacy(sparkSession)) &&
         (dbName.equals("default") || databaseLocation.endsWith(".db"))) {
       val carbonStorePath = CarbonProperties.getStorePath()
-      val hiveStorePath = sparkSession.conf.get("spark.sql.warehouse.dir", 
carbonStorePath)
+      val hiveStorePath = sparkSession.conf.get("spark.sql.warehouse.dir")
       // if carbon.store does not point to spark.sql.warehouse.dir then follow 
the old table path
       // format
       if (carbonStorePath != null && !hiveStorePath.equals(carbonStorePath)) {

Review comment:
       In my case, a vanilla apache version of spark on yarn, the default 
warehouse location is under the local working directory where the thrift server 
launch. 
   
   If you check spark code, the default value of `spark.sql.warehouse.dir` 
barely comes from Java URI implementation.
   
   
https://github.com/apache/spark/blob/6c792a79c10e7b01bd040ef14c848a2a2378e28c/sql/catalyst/src/main/scala/org/apache/spark/sql/internal/StaticSQLConf.scala#L33-L37
   
   
https://github.com/apache/spark/blob/47dc332258bec20c460f666de50d9a8c5c0fbc0a/core/src/main/scala/org/apache/spark/util/Utils.scala#L1976
   
   
![image](https://user-images.githubusercontent.com/3809732/83354246-fab16c00-a389-11ea-96ca-48e28ebe799c.png)
   
   
   the problem of PR #3675 is : 
   1. carbon store path is not taken effect, default database warehouse URI 
(local fs, file://...) is used. <---  this pr fix this
   2. the scheme of default database warehouse URI is removed by 
`FileFactiry.getUpdatedFilePath` in `CarbonEnv.getDatabaseLocation`. And this 
is not an absolute URI string, so spark will make 
    qualified path in `SessionCatalog#createTable`. <--  #3675 fix this




----------------------------------------------------------------
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]


Reply via email to