On Wednesday 11 November 2009 07:15:36 am Josh Boyer wrote:
On Tue, Nov 10, 2009 at 11:24:50PM -0800, Jitesh Shah wrote:
>So, I picked up the sign_unsigned.py script from releng. I replaced the
> keys in there with our keys, tweaked some minor stuff here and there and
> managed to get it running. I use it as
>"./sign_unsigned.py --level <level> <tag-name>"
>and it runs alright. I can see that the signatures are cached under the
> sigcache directory (but NOT embedded in the rpms themselves, which makes
> sense since the rpm can probably be a part of different tags and might be
> signed differently within each tag)
>
>So, I thought, well, mash would be the one which'll embed the keys in the
> rpms. So, I set strict_keys to True.. added my key to the keys list in my
> .mash file. mash has no problems with the rpms and it can verify the
> signatures alright. But, it still doesn't embed the signatures in the rpm
> (is it supposed to?). So, the created repository still has all rpms
> unsigned.
>
>What am I missing here? where to the rpms get signed actually?
The sign_unsigned script should eventually do a koji API call to do
'write-signed-rpm' on the packages you are signing. That will assemble
signed RPMs in koji itself, which mash will download and used.
Fedora Rel-Eng doesn't use sign_unsigned anymore because we have a signing
server setup now. However, it should still work.
it still works. EPEL releng still
uses it. you need to make sure to add --
write-rpms to you command. the signed rpms will then get written.
Dennis