Datasealer key rotation and CAS

[Cantor, Scott]

On 10/27/20, 1:57 PM, "users on behalf of Paul B. Henson" <users-bounces at shibboleth.net on behalf of henson at cpp.edu> wrote:

> Oh, okay; I misunderstood the purpose of the refresh interval. I had assumed it was how often it looked to see if the
> keystore had changed and reload the keys within. But it seems its purpose is only to check and see if the default key has
> changed so it could start using it to encrypt stuff.

Also to generally optimize under the presumption that all the nodes are usually on the same default key, but mainly that; the default key is the one it encrypts with itself. I was concerned I was missing something with the CAS usage but I didn't think so.

> So as long as the keystore is updated on all of the nodes at the same time, while each node might start using the newest
> key for encryption purposes at a different time, all of them will be able to decrypt stuff encrypted by other nodes with
> said key.

Yes, that's the only critical assumption that can't be (horribly) violated.

-- Scott