taylorobyen commented on PR #10613: URL: https://github.com/apache/nifi/pull/10613#issuecomment-3691360738
If it is intended behavior to parse the timestamp differently based off the input, I think the reader controller services should have their descriptions updated. In [NIFI-12710](https://issues.apache.org/jira/browse/NIFI-12710), it modifies the parsing under the hood, but nothing was updated description wise to indicate to the end user that this behavior has changed. When I upgraded my production instances of NiFi from 1.20.0 to NiFi 2.5.0 and noticed that my timestamps were now being parsed incorrectly the first place I checked was the JsonTreeReader timestamp field to see if parsing had changed. It wasn't until I checked out the project and viewed the source code that I found that I need to provide my milliseconds as integers to have my timestamps parsed correctly. The current description from JsonTreeReader only mentions milliseconds (other readers such as CSV also have the same description): <img width="1500" height="1206" alt="image" src="https://github.com/user-attachments/assets/02a8fd09-ec25-4b0c-9d4c-5b638055f688"; /> My use case for these timestamps is that I load JSON payloads into my PostgreSQL database which have timestamp fields generated in Python i.e: ```python ms = int(time.time() * 1000) ``` Which will generate something like `1766663148057.1487`. To work around the current behavior, I'm rounding down to an integer so that the timestamps are parsed correctly. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
