The benefit of the solution I mention is simply that it can be implemented 
without changing Kafka, and I provided it more as a side note for people 
reading this list that might not have time to wait for this KIP to land into a 
released version. I do think that the KIP proposal would be very useful.

Cheers,
noa

> On 4 Dec 2020, at 19:02, Boyang Chen <reluctanthero...@gmail.com> wrote:
> 
> Thanks Noa for the suggested path. Like you mentioned, I feel this
> mechanism is a little bit overkill for a simple security file reloading
> case. Could you provide more context on the benefit of doing a customized
> KeyManager setup? TBH, I don't see Kafka going deep into these low level
> security details yet.
> 
> Best,
> Boyang
> 
> On Fri, Dec 4, 2020 at 4:02 AM Noa Resare <n...@resare.com> wrote:
> 
>> Hi Boyang,
>> 
>> I think that it would improve the ergonomics of dealing with short lived
>> certificates to have this be the default behaviour.
>> 
>> It should be noted that transparently reloading certificates and keys when
>> they changed on disk can be implemented right now registering a custom
>> KeyManagerFactory, but to say that the JDK is designed to make this easy
>> would be an overstatement. The things that we do to get this working:
>> 
>> 1. Create a class implementing SecurityProviderCreator that will return a
>> Provider that registers a custom KeyManagerFactory implementation.
>> 2. This custom KeyManagerFactory would return KeyManager instances that
>> implements X509ExendedKeyManager
>> 3. The custom KeyManager would return cached but up to date values for the
>> getCertificateChain() and getPrivateKey() methods.
>> 5. Configure Kafka with security.providers referencing the class defined
>> in 1)
>> 
>> This is not something I would wish upon anyone, but it works. Solving this
>> for everyone inside Apache Kafka seems like a much preferred solution.
>> 
>> Cheers
>> noa
>> 
>> ps. It seems my apple.com <http://apple.com/> email address ends up on
>> the list as apple.com <http://apple.com/>.INVALID. Is this a known
>> problem? For now I’m working around it by using my personal email.
>> 
>>> On 4 Dec 2020, at 01:28, Boyang Chen <reluctanthero...@gmail.com> wrote:
>>> 
>>> Hey there,
>>> 
>>> I would like to start the discussion thread for KIP-687:
>>> 
>> https://cwiki.apache.org/confluence/display/KAFKA/KIP-687%3A+Automatic+Reloading+of+Security+Store
>>> 
>>> This KIP is trying to deprecate the AlterConfigs API support of updating
>>> the security store by reloading path in-place, and replace with a
>>> file-watch mechanism inside the broker. Let me know what you think.
>>> 
>>> Best,
>>> Boyang
>> 
>> 

Reply via email to