Hello,
Thanks.
In fact I use HDP 2.6.5 and previous Orc version with transactionid for example
and the update flag.
Sorry with the row__id iw would have been easier
So, Here after the Orc files content (with hive --orcfiledump)
hive --orcfiledump hdfs:///delta_0198994_0198994_/bucket_000
Hi David,
There is no tombstone for the updated record.
In ACID v2 there is no update for the rows. Only insert and delete. So update
is handled as delete (old) row, insert (new/independent) row.
The delete is stored in the delete delta directories., and the file do not have
to contain the {row}
Hi,
It works pretty well but... still problems sometimes occur
Do we have to separate operations ?
Here after Orc files content:
hive --orcfiledump hdfs:///delta_0198994_0198994_/bucket_0
{"operation":0,"originalTransaction":198994,"bucket":0,"rowId":14,"currentTransaction":198994,"r
