[
https://issues.apache.org/jira/browse/NIFI-5154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16466168#comment-16466168
]
ASF GitHub Bot commented on NIFI-5154:
--------------------------------------
Github user mcgilman commented on the issue:
https://github.com/apache/nifi/pull/2678
Thanks for the PR @markap14! I ran into an issue with the proposed
solution. Steps to replicate:
- Create a PG
- Add a Processor (InvokeHTTP) to the PG.
- Add a CS (SSLContextService) to the PG.
- Reference the CS in the Processor.
- Copy the Processor out of the PG.
- Processor CS reference is no longer valid and it's no longer a
referencing component of the CS. Yay!
- Create another PG at the same level as the in Step 1.
- Move the copied Processor into the new PG.
- Move the new PG inside the PG created in Step 1.
- Copied Processor CS reference is valid again. Yay!
- However, it does not appear as a referencing component of the CS :/
> Out of Scope processors can block Controller Services
> -----------------------------------------------------
>
> Key: NIFI-5154
> URL: https://issues.apache.org/jira/browse/NIFI-5154
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework
> Reporter: Mark Payne
> Assignee: Mark Payne
> Priority: Major
> Fix For: 1.7.0
>
>
> On root canvas level PG1 and PG2 are created. User1 and user2 have access to
> PG1 while only user2 has access to PG2. User1 or user 2 creates a processor
> and corresponding CS in PG1. At this time both user1 and user2 can disable
> and modify that CS. User2 then copies the processor referencing that CS in
> PG1. That snippet is then pasted inside of PG2. Since PG2 is not a sub
> process group of PG1 the CS referenced in that copied snippet by UUID is out
> of scope for that pasted processor. The CS in PG1 still sees that referencing
> processor from PG2 and now user1 can no longer disable and modify the CS in
> PG1.
> The pasted processor is clearly out of scope of referenced CS. The processor
> would still present as invalid when pasted and still reference the CS's UUID
> in its snippet until property was updated or until processor was moved to a
> new location that was within scope of the CS. The fix here would be to make
> sure the CS does not reference any processors that are out of scope. So in
> this specific scenario it would not block.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
