1 |
On Saturday 27 March 2004 04:28, Andrew Cowie wrote: |
2 |
> On Wed, 2004-03-24 at 16:07, Chris Bainbridge wrote: |
3 |
> > c) for each signature in .secure/*.asc check whether its in the ACL |
4 |
> > list, then call `gpg --verify .secure/sig.asc .secure/hash` to verify it. |
5 |
> > We can set auto-key-retrieve in case we don't already have the key. |
6 |
> |
7 |
> Something that I've been trying to figure out in this whole discussion |
8 |
> of rapidly expiring keys is what happens to machines that don't have |
9 |
> at-will access to the public internet: |
10 |
> |
11 |
> .. a disconnected machine (like a laptop) who is away from the internet |
12 |
> for days or weeks at a time, or |
13 |
|
14 |
At the moment of syncing a timestamp is stored which is used for determining |
15 |
the validity of the key. The condition for this to work is that the local |
16 |
machine is not compromised. But if it were, all odds were of anyway. |
17 |
|
18 |
> .. a server node that doesn't get its packages from the net at all, but |
19 |
> rather is part of a production farm which gets its updates from some |
20 |
> local local mirror/build machine only when the site administrators make |
21 |
> a new local set of packages available to that server farm. |
22 |
|
23 |
The timestamp needs to be made available to the client machines, or (more |
24 |
likely) the server provides it's own list of allowed keys possibly including |
25 |
local administrators' keys. A local list requires a configuration option that |
26 |
specifies an alternative key that can override the gentoo provided |
27 |
signatures. |
28 |
|
29 |
> What happens in those scenarios? |
30 |
|
31 |
See above |
32 |
|
33 |
Paul |
34 |
|
35 |
-- |
36 |
Paul de Vrieze |
37 |
Gentoo Developer |
38 |
Mail: pauldv@g.o |
39 |
Homepage: http://www.devrieze.net |