simonbence commented on PR #7965: URL: https://github.com/apache/nifi/pull/7965#issuecomment-1790512269
@pvillard31 @exceptionfactory Thanks for the feedback! As of this given case, the behaviour is specific to the way this particular implementation works. Other implementations, if they are caching some internal state depending on "controller service" property might have a similar challenge. @exceptionfactory You are perfectly right, the Flow Registry Client currently has no such lifecycle states like active-deactivated, that is correct. At the time of changing the Client to a component, this was an intentional choice due to the number of questions would come with this behavioural change (and due to avoiding any break in the previous behaviour). The questions include: what should happen with the existing versioned flows when the registry is deactivated? It looks like a different case compared to the situation when the registry is unintenionally unreachable and might result some confusion. An other possibility you mention is to "hook on" some existing event like validation or as you mention, the (de)activation of the dependent Controller Services but I am not aware of any support in the framework for this. What I propose to do is to move on with this fix (which in some cases can cause misbehaviour as it is now) and for a wider scope solution it would be beneficial to start a discussion about the level of support we aim for: I can easily imagine that an additional hookpoint might be useful for other components as well. -- 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]
