Manikumar, Satish. Thanks for the review! As I understand, you are not in
favor of this KIP, and I do agree that having a pluggable mechanism for
sensitive data / metadata is preferable/more future-proof.

On Wed, Dec 12, 2018 at 8:12 AM Satish Duggana <satish.dugg...@gmail.com>
wrote:

> Agree with Manikumar on having pluggable mechanism for entities
> required/created for delegation token mechanism. I will cover that as
> part of KAFKA-7694.
>
> Thanks,
> Satish.
> On Tue, Dec 11, 2018 at 12:35 PM Manikumar <manikumar.re...@gmail.com>
> wrote:
> >
> > Hi,
> >
> > Thanks for the KIP.
> >
> > Currently, master/secret key is stored as plain text in server.properties
> > config file.
> > Using master secret key as shared secret is again a security risk. We
> have
> > raised KAFKA-7694
> > to implement a ZooKeeper based master/secret key management to automate
> > secret key rotation.
> >
> > As you mentioned in the alternatives sections, it is good to have
> pluggable
> > mechanism for
> > token storage and master key generation. We can implement pluggable
> > interfaces for token storage
> > and master key generation as part of KAFKA-7694. This will provide us out
> > of the box implementation
> > using ZooKeeper and pluggable interfaces for custom implementations.
> >
> > What do you think?
> >
> > Thanks,
> > Manikumar
> >
> > On Sat, Dec 1, 2018 at 9:37 PM Attila Sasvári <asasv...@apache.org>
> wrote:
> >
> > > Hi All,
> > >
> > > I have a proposal to allow Kafka brokers to encrypt sensitive metadata
> > > information about delegation tokens.
> > >
> > > As of now, delegation token metadata is stored in an unencrypted
> format in
> > > Zookeeper. Having the possibility to encrypt-then-MAC token information
> > > would be beneficial in Kafka installations where Zookeeper is not on a
> > > private network.
> > >
> > > Please take a look at
> > >
> > >
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-395%3A+Encypt-then-MAC+Delegation+token+metadata
> > > and let me know what you think.
> > >
> > > - Attila
> > >
>

Reply via email to