[GitHub] [incubator-hudi] afilipchik commented on a change in pull request #1516: [HUDI-784] Adressing issue with log reader on GCS

[GitBox <gi...@apache.org>]

afilipchik commented on a change in pull request #1516:
URL: https://github.com/apache/incubator-hudi/pull/1516#discussion_r411560192



##########
File path: hudi-common/src/main/java/org/apache/hudi/common/table/log/HoodieLogFileReader.java
##########
@@ -79,6 +79,7 @@
       this.inputStream = fsDataInputStream;
     }
 
+    fsDataInputStream.seek(0);

Review comment:
       magicBuffer check was failing on file open, as it couldn't find a beginning of the HUDI block in the log file. 
   That was throwing exception that was killing compaction. In the debug, it appeared that content of magicBuffer was incorrect, and steam offsets we off. Only happened when more that 1 files was scheduled to be processed. So, I didn't test all the variations originally (non static magicBuffer without seek and seek with static) as was in hurry to fix, so not sure which one actually fixes the issue. Agree that seek(0) is weird and probably non static is a fix. Added comment and additional check..




----------------------------------------------------------------
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:
users@infra.apache.org