You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Koji Kawamura (JIRA)" <ji...@apache.org> on 2018/01/09 08:52:00 UTC

[jira] [Created] (NIFI-4752) REPLAY events returned by WriteAheadProvenanceRepository have child FlowFile UUID as event FlowFile UUID

Koji Kawamura created NIFI-4752:
-----------------------------------

             Summary: REPLAY events returned by WriteAheadProvenanceRepository have child FlowFile UUID as event FlowFile UUID
                 Key: NIFI-4752
                 URL: https://issues.apache.org/jira/browse/NIFI-4752
             Project: Apache NiFi
          Issue Type: Bug
          Components: Core Framework
    Affects Versions: 1.4.0
            Reporter: Koji Kawamura


The ['Provenance Events' documentation|https://nifi.apache.org/docs/nifi-docs/html/developer-guide.html#provenance_events] describes about REPLAY event as follows:
{quate}
Indicates a provenance event for replaying a FlowFile. The UUID of the event indicates the UUID of the original FlowFile that is being replayed. The event contains one Parent UUID that is also the UUID of the FlowFile that is being replayed and one Child UUID that is the UUID of the a newly created FlowFile that will be re-queued for processing
{quate}

The default PersistentProvenanceRepository behaves as written in the doc. But WriteAheadProvenanceRepository returns REPLAY events having Child UUID as its FlowFile UUID instead.

Here is the lines of code that set FlowFile UUID for the provenance events.
https://github.com/apache/nifi/blob/master/nifi-nar-bundles/nifi-provenance-repository-bundle/nifi-persistent-provenance-repository/src/main/java/org/apache/nifi/provenance/schema/LookupTableEventRecord.java#L276-L280

WriteAheadProvenanceRepository does not seem to have 'FlowFile UUID'  value persisted, which is set by FlowController when replay events are registered. Instead, WriteAheadProvenanceRepository fill 'FlowFile UUID' from updated or previous 'UUID' attribute.
I don't know much background on why it is implemented this way, but it seems it drops 'FlowFile UUID' to reduce IO based on an assumption that it can be set by attributes.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)