F11: things to check in a release tree before a release

[Seth Vidal]

On Wed, 19 Nov 2008, Daniel P. Berrange wrote:

> On Wed, Nov 19, 2008 at 03:52:15PM -0500, Seth Vidal wrote:
>> Going through a variety of checks for things that could be broken in a
>> release tree, I decided to type up a list of the things we should be
>> checking for.
>>
>> ??? metadata matches
>> ??? comps correct
>> ??? file conflicts
>> ??? normal conflicts
>> ??? no obsoleted pkgs in tree
>> ??? report all triggers
>> ??? look for all postrans/pretrans things
>> ??? unresolveable requires
>> ??? self-provided filedeps
>> ??? self-provided normal deps
>> ??? old versions of pkgs left over
>>
>>
>> Some of them are not explicitly things which are broken in a release, just
>> things it would be wise to make sure are sane. Triggers,pretrans and
>> posttrans are good examples of that.
>>
>> Can anyone think of any others that we should report/check for?
>
> All files in the .treeinfo file actually exist, and that the .treeinfo
> has entries for baremetal & xen kernels (xen may or may not point to
> the same kernels as baremetal). We now rely on .treeinfo files for
> virtualization provisioning needs, so brokenness is a show-stopper.


verifytree already does that. but yes, correct.

-sv