Re: [netconf] ietf crypto types - permanently hidden

[Kent Watsen <[e-mail]>]

Hi Juergen,

> Some comments:
> 
> - Not sure what security best practice is and where it is defined.

https://mailarchive.ietf.org/arch/msg/netconf/duOgrXt6P-WwO-Vdole9SsxS2vg <https://mailarchive.ietf.org/arch/msg/netconf/duOgrXt6P-WwO-Vdole9SsxS2vg>

>  Creating the key on the device has the benefit that the key may
>  never have to leave the device but it also requires that I can trust
>  the device to create good keys. (People recall the glitch that led
>  to weak openssh keys on many Linux systems? And given todays world,
>  which devices should I trust to generate strong keys?) I might
>  decide that creating keys with known software I am willing to trust
>  is far better than trusting a key generator shipped by a random
>  device manufacturer.

Yes, and I've said the same here as well.  Both options should be supported.

> - Side effects are bad and should be avoided. If something requires an
>  rpc/action, make it an rpc/action.
> 
> - If you create something, give it a name; once things are named (or
>  have other unique properties), config can then refer to these named
>  objects. A create key operation might simply return a fingerprint
>  for the key as the key's name.

I'm unsure how to apply this beyond what has been discussed.


Kent