Re: [DISCUSSION] KIP-421: Support resolving externalized secrets in AbstractConfig

2019-03-21 Thread Tejal Adsul
I have addressed the comments 1 and 2 in the KIP. 3. The example is a bit misleading with the password in it. I have modified it. We basically wanted to show that you cam pass any additional parameters required by the config provider 4. Yes all the public config classes (ProducerConfig, Consumer

Re: [DISCUSSION] KIP-421: Support resolving externalized secrets in AbstractConfig

2019-03-21 Thread Tejal Adsul
@Colin we won’t be supporting the subscriber mode currently and it will be added as a future work 2. By disabling the feature the constructor will work as it earlier. If we know the configs don’t have any indirect values or we want the indirect values to remain unresolved we will can just do so

Re: [DISCUSSION] KIP-421: Support resolving externalized secrets in AbstractConfig

2019-03-14 Thread Colin McCabe
Hi Tejal, Thanks for the update. One of the critical parts of the ConfigProvider interface is the ability to monitor changes to a configuration key through ConfigProvider#subscribe and ConfigProvider#unsubscribe, etc. I don't see how the proposed API supports this. Can you clarify? Also, it

Re: [DISCUSSION] KIP-421: Support resolving externalized secrets in AbstractConfig

2019-03-14 Thread Rajini Sivaram
Hi Tejal, Thanks for the updates. A few comments: 1. In the standard KIP template, we have two sections `Public Interfaces` and `Proposed Changes`. Can you split the section `Proposal` into two so that public interface changes are more obvious? 2. Under `Public Interfaces`, can you s

Re: [DISCUSSION] KIP-421: Support resolving externalized secrets in AbstractConfig

2019-03-11 Thread Tejal Adsul
Hi Folks, I have accommodated most of the review comments for https://cwiki.apache.org/confluence/display/KAFKA/KIP-421%3A+Support+resolving+externalized+secrets+in+AbstractConfig . Reopening the thread for further discussion. Please let me know your thoughts on it. Thanks, Tejal On 2019/01/2