You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When sealed secret was first designed, it was declared to support integrity protection via signature. Currently, sealed secret does not support this ability, and a fakesignature is used to replace the original valid JWT signature. This issue mainly hopes to solve several issues about signatures:
Add support for CDH to verify the signature of a signed sealed secret. The public key can either be accessed from KBS, or be set via CDH's config file.
Add support for sealed secret client tool to generate a signed sealed secret.
This issue is purely from technical point view, but we can start from this point to talk more about the signed sealed secret use case.
The text was updated successfully, but these errors were encountered:
Where should we get the key to verify the secret? Can we expect the kid field to have a resource URI? Does that make sense even for secrets that point to some other HSM?
Resource URI should be fine. And to avoid circular dependency, HSM can play a role as a storage repository. This means the sealed secret's key cannot be wrapped in another sealed secret.
Other choices are a path inside guest image, a key from initdata. The integrity of both can be protected by RA.
Ok I think getting from the KBS is fine although initdata support would also be good.
This is going to impact the user experience a bit. Now people will have to provision more stuff. I wonder if it makes sense to support unsigned secrets as well and if there is any way to do that safely.
When sealed secret was first designed, it was declared to support integrity protection via signature. Currently, sealed secret does not support this ability, and a
fakesignature
is used to replace the original valid JWT signature. This issue mainly hopes to solve several issues about signatures:This issue is purely from technical point view, but we can start from this point to talk more about the signed sealed secret use case.
The text was updated successfully, but these errors were encountered: