[GitHub] [carbondata] junyan-zg commented on a change in pull request #3186: [CARBONDATA-3345]A growing streaming ROW_V1 carbondata file would be ingored some InputSplits

[GitBox]

junyan-zg commented on a change in pull request #3186: [CARBONDATA-3345]A growing streaming ROW_V1 carbondata file would be ingored some InputSplits
URL: https://github.com/apache/carbondata/pull/3186#discussion_r278389589
 
 

 ##########
 File path: integration/presto/src/main/java/org/apache/carbondata/presto/impl/CarbonTableReader.java
 ##########
 @@ -291,7 +292,12 @@ private CarbonTableCacheModel getValidCacheBySchemaTableName(SchemaTableName sch
         // Use block distribution
         List<List<CarbonLocalInputSplit>> inputSplits = new ArrayList(
             result.stream().map(x -> (CarbonLocalInputSplit) x).collect(Collectors.groupingBy(
-                carbonInput -> carbonInput.getSegmentId().concat(carbonInput.getPath()))).values());
+                carbonInput -> {
+                  if (FileFormat.ROW_V1.equals(carbonInput.getFileFormat())){
+                    return carbonInput.getSegmentId().concat(carbonInput.getPath()).concat(carbonInput.getStart() + "");
 
 Review comment:
   Only for the map key，without 'carbonInput.getStart()'，and only 'carbonInput.getSegmentId().concat(carbonInput.getPath())',  the some small files would be regarded as the same file  and it would be ingore.
   And  I think the bug doesn't cause in COLUMN_V3,  the COLUMN_V3 status contains compacted and success, however it queryed normally, so I only changed the ROW_V1 format.

----------------------------------------------------------------
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:
[hidden email]


With regards,
Apache Git Services