> Andy LoPresto
> alopre...@apache.org
> alopresto.apa...@gmail.com

Thanks for the reply. I tried the steps you lined but but the only feasible 
option is as lined out below, feature parity with the standalone mode.

> The quick answer is that there is currently a feature availability 
> difference between the standalone mode and the client/server mode of 
> the TLS Toolkit. The behavior you are requesting is currently in the 
> standalone mode but not in the client/server mode. This is due to a 
> difference in their implementation logic, which is being addressed as 
> part of a larger refactoring under NIFI-5462 [1], which is a component 
> of the TLS Toolkit epic NIFI-5458 [2]. 

Is this in progress? I am happy to pick this up and do it per guidance. Its an 
important feature.

> Example keytool output for keystore with external CA public certificate 
> in chain:  
  
> Owner: CN=intermediate_signed.nifi.apache.org, OU=NIFI
> Issuer: CN=Intermediate CA

> Owner: CN=Intermediate CA
> Issuer: CN=Root CA

In my experience, getting a Sub-CA in an enterprise organization is a challenge 
and it exposes other risks. Its easier for me to code up a solution for 
everyone. :)

> The NiFi TLS Toolkit was designed to facilitate the deployment and 
> configuration of TLS behavior _in an environment lacking an enterprise 
> security team_.

I wish an enterprise was that simple. Many times people need to take ownership 
themselves. Thats why offer to code up feature parity.

> I am sorry the PKIX error message is unclear. The error is a 
> Java-native exception and we can make an effort to capture it and 
> provide a more user-friendly message in the future. However, searching 
> "PKIX path building failed nifi” [7] does return a number of helpful 
> articles and Q/A posts with instructions to resolve the issue. 

I usually catch the error, add a little more content to the error message, and 
re-throw it. It does give the bes of both worlds.

> I encourage you to share further thoughts on concrete steps to improve 
> this process that do not open users up to unseen security risks, and to 
> follow the Jiras discussed above for further developments in improving 
> the behavior and ease-of-use of the TLS Toolkit. 

I agree. If there is no one working on
  - https://issues.apache.org/jira/browse/NIFI-5460

I will happily help and coordinate with whomever is working on 
  - https://issues.apache.org/jira/browse/NIFI-5462

Erik Anderson
Bloomberg
https://www.linkedin.com/in/erikanderson/

Reply via email to