You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@parquet.apache.org by shimin yang <ys...@gmail.com> on 2019/06/27 09:11:36 UTC

Close parquet writer twice ends up with invalid state

Hi all,

When I try to close ParquetWriter, the HDFS throws Exception "Uable to
create new block". Then I try to close it again, the ParquetWriter throws

java.io.IOException: The file being written is in an invalid state.

I look at the code, in the InternalParquetRecordWriter#
flushRowGroupToStore, the state will turn to COLUMN and will not go back to
BLOCK if any exception was thrown by FSOutputStream.

Is there any way that I can close it twice? Since I think the data is still
in the memory or local files.

Best,
Shimin Yang