On Wed, 2012-05-02 at 20:06 +0200, Jakub Hrozek wrote:
On Wed, May 02, 2012 at 01:28:42PM -0400, Stephen Gallagher wrote:
> There was a bug in finish_upgrade() where it would return EOK if
> it succeeded in canceling the transaction due to an error. We
> should instead be returning the original error.
> I noticed this while reviewing Jan's ghost user patches. This may also
> explain a strange bug a user saw yesterday in #sssd when upgrading from
> 1.5.1 to 1.8.0. (Or at least it would help it fail more clearly).
Pushed to master.