[ 
https://issues.apache.org/jira/browse/BEAM-6292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16797018#comment-16797018
 ] 

Ismaël Mejía commented on BEAM-6292:
------------------------------------

Thanks for the explanation [~mblmat]. I see more clearly the case now and it is 
interesting the KMS/IAM approach so it can only be decrypted by the valid. I 
still have the doubt then for all other IOs? I mean I don't see similar methods 
in other IOs which may be visible in the staging location so if the problem is 
as you mention this deserves a global solution, not per IO, isn't there other 
way to achieve this? (Also now I remember that the google guys added Kms based 
authentication recently).

I am moving this subject to the mailing list because I really don't know about 
Google's recommendation for this case or maybe someone else can have an idea.

In the meantime I will do a PR to remove the extra `withEncryptedPassword` 
method and rely on decryption of Password only if the PasswordDecrypter is 
provided to make the API simpler.

> PasswordDecrypter: Delay decryption / Avoid serialization
> ---------------------------------------------------------
>
>                 Key: BEAM-6292
>                 URL: https://issues.apache.org/jira/browse/BEAM-6292
>             Project: Beam
>          Issue Type: Improvement
>          Components: io-java-cassandra
>            Reporter: Mathieu Blanchard
>            Assignee: Mathieu Blanchard
>            Priority: Minor
>              Labels: triaged
>             Fix For: 2.12.0
>
>          Time Spent: 10h 40m
>  Remaining Estimate: 0h
>
> Currently, the password is decrypted before the serialization of the pipeline 
> and this causes the raw version to be visible to everyone on the staging 
> location.
> To avoid this, we delayed the decryption of the password when connecting to 
> the cluster, which ensures that the raw password is never serialized in the 
> pipeline.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to