[GitHub] [hudi] HEPBO3AH commented on issue #6212: [SUPPORT] Hudi creates duplicate, redundant file during clustering

Tue, 02 Aug 2022 14:52:37 -0700 


HEPBO3AH commented on issue #6212:
URL: https://github.com/apache/hudi/issues/6212#issuecomment-1203249824

   Hi,
   
   Here is the code sample to replicate it:
   
   ```
       val spark = SparkSession
         .builder()
         .master("local[3]")
         .config("spark.serializer", 
"org.apache.spark.serializer.KryoSerializer")
         .config("spark.hadoop.fs.s3a.aws.credentials.provider", 
"com.amazonaws.auth.profile.ProfileCredentialsProvider")
         .config("spark.eventLog.enabled", "true")
         .config("spark.eventLog.dir", logPath)
         .getOrCreate();
   
       import spark.implicits._
   
       val ids = 1 to 5000 grouped(1000) toSeq
   
       for (idSection <- ids) {
   
         val df = idSection.toDF("id")
   
         df
           .write
           .format("org.apache.hudi")
           .option(TBL_NAME.key(), tableName)
           .option(TABLE_TYPE.key(), COW_TABLE_TYPE_OPT_VAL)
           .option(RECORDKEY_FIELD.key(), "id")
           .option(PARTITIONPATH_FIELD.key(), "")
           .option(OPERATION.key(),  INSERT_OPERATION_OPT_VAL)
           .option(INLINE_CLUSTERING.key(), "true")
           .option(INLINE_CLUSTERING_MAX_COMMITS.key(), "1")
   
           .mode(SaveMode.Append)
           .save(path)
       }
   ```
   
   What we see in `.hoodie`:
   
   <img width="994" alt="image" 
src="https://user-images.githubusercontent.com/15118722/182479505-e5567179-6f80-4e7d-88f9-e463458b9a06.png";>
   
   The files that were double created and then deleted. The delete markers 
correspond to the the clustering commit ids:
   
   <img width="809" alt="image" 
src="https://user-images.githubusercontent.com/15118722/182479683-f1c95881-b0ba-4a62-845e-c4d0390a3483.png";>
   
   These have 1 to 1 relationship with clustering. If you set clustering to 
happen every 2 commits, only 2 delete markers will be present - one for each 
clustering as we do 5 passes in total in the code sample.
   


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to