[
https://issues.apache.org/jira/browse/HIVE-21146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16943578#comment-16943578
]
David Lavati commented on HIVE-21146:
-------------------------------------
All test failures are due to this new validation step, it seems like the scope
of affected filesystems has to be limited. I've yet to find a proper solution
for this, as e.g. AzureBlobFileSystem and its store are simply extending their
base classes.
> Enforce TransactionBatch size=1 for blob stores
> -----------------------------------------------
>
> Key: HIVE-21146
> URL: https://issues.apache.org/jira/browse/HIVE-21146
> Project: Hive
> Issue Type: Bug
> Components: Streaming, Transactions
> Affects Versions: 3.0.0
> Reporter: Eugene Koifman
> Assignee: David Lavati
> Priority: Major
> Labels: pull-request-available
> Attachments: HIVE-21146.patch
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Streaming Ingest API supports a concept of {{TransactionBatch}} where N
> transactions can be opened at once and the data in all of them will be
> written to the same delta_x_y directory where each transaction in the batch
> can be committed/aborted independently. The implementation relies on
> {{FSDataOutputStream.hflush()}} (called from OrcRecordUpdater}} which is
> available on HDFS but is often implemented as no-op in Blob store backed
> {{FileSystem}} objects.
> Need to add a check to {{HiveStreamingConnection()}} constructor to raise an
> error if {{builder.transactionBatchSize > 1}} and the target table/partitions
> are backed by something that doesn't support {{hflush()}}.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
