Hi.  Thanks for the detailed description.  Was able to reproduce this with
the current code, and am working on a candidate fix.  I've entered
NIFI-10851 to track.

- https://issues.apache.org/jira/browse/NIFI-10851


On Tue, Nov 15, 2022 at 11:53 AM John McGinn via users <
[email protected]> wrote:

>
> I'm using NiFi 1.18.0, and I use the controller services configure, enable or 
> disable screen, to determine what processors is using it. That way I can 
> determine if I can delete the controller service or not. Or, if I have to 
> migrate from 1 controller service to a controller service in a processor 
> group. (For instance, my flow gets too big, so I move the processors to a 
> group, but the controller services stay at the root level.) (That way I can 
> know for sure a controller service can get deleted because nothing is using 
> it, rather than deleting the controller service and having to find all 
> processors after the fact that are now invalid because of that.)
>
>
> To test this, I was able to do the following. Create a processor group, add 
> ConvertRecord in the group, and choose a standard AvroReader as the service 
> at the root level. Name all of them for uniqueness. Copy the processor group, 
> go into the ConvertRecord and choose a new processor group level AvroReader, 
> and name these items uniquely as well. Go back to the root level and look at 
> the root AvroReader, and it was showing both ConvertRecords, even the one in 
> the copied group that has a new AvroReader in it.
>
>
> I just tried that test in a 1.9.0 NiFi instance, and it didn't have the same 
> problem.
>
> John McGinn
>

Reply via email to