[
https://issues.apache.org/jira/browse/NIFI-6244?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Matt Burgess reopened NIFI-6244:
--------------------------------
Reopening due to the possibility of duplicate and/or missing rows. We won't be
able to use newid() on multiple fetches due to the fact that each partition
fetch may sort the rows arbitrarily, so some rows may show up in multiple
partitions and some rows may not show up in any partition.
I will investigate picking a column from the table (perhaps always the first
one, or possibly finding one that definitely can be ordered) to order by. If
that is not viable, the original PR should be reverted and this issue can be
closed.
> Allow GenerateTableFetch to partition without a max-value column
> ----------------------------------------------------------------
>
> Key: NIFI-6244
> URL: https://issues.apache.org/jira/browse/NIFI-6244
> Project: Apache NiFi
> Issue Type: Bug
> Reporter: Matt Burgess
> Assignee: Matt Burgess
> Priority: Major
> Fix For: 1.10.0
>
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> For some RDBMS systems / SQL dialects, it is not valid to try and
> paginate/partition rows without an ORDER BY clause. MS SQL Server 2012+ is
> one example (see
> [here|https://stackoverflow.com/questions/49262675/issue-with-generate-table-fetch-sql-server-2016]
> for more details). However some systems/dialects allow for a "dummy" or
> default ordering, which can be generated when no ordering clause is provided
> to the database adapter. Again with MS SQL Server 2012+ as an example, you
> can ORDER BY newid().
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
